#!/usr/bin/env Rscript # This script will be used to perform data analysis and create display items from qBiCo indices. #Raw qPCR data to qBiCo indices#### {#dependencies#### # install.packages("xlsx") library(xlsx) library(plyr) library(dplyr) library(hash) library(zeallot) library(gtable) library(gridExtra) library(grid) library(gridtext) library(rio) library(tidyr) library(ggplot2) library(fANCOVA) library(ggpubr) library(naniar) library(writexl) } #From Excel file values to create a standard curve and obtain the amount of copy numbers and Cq standard deviations. raw_to_parameters <- function(raw_data, manual_exlusion=FALSE,standards_list) { #if manual exlusion is TRUE, then ONLY exclude the ones that were manually provided. if(isTRUE(manual_exlusion)){ raw_data$Cq[raw_data$excluded == 1] <- NA } #Create Standards and Samples Cq data frames df_Cq_avg <- raw_data %>% spread(Target, Cq) df_Cq_avg <- subset(df_Cq_avg, !grepl('NA',df_Cq_avg$Sample) & !grepl('NTC',df_Cq_avg$Sample)) df_Cq_avg <- subset(df_Cq_avg, select = -c(Fluor,DNA_input_ng,Elution_volume_ul,excluded)) if ('NA' %in% colnames(df_Cq_avg)){ colnames1 <- c('Well','Sample','Cq_Converted','Cq_Genomic', 'Cq_Long', 'Cq_IPC','Cq_Short') colnames2 <- c('Well','Sample','Cq_Converted','Cq_Genomic','Cq_Long', 'Cq_IPC','Cq_Short','SQ_Converted','SQ_Genomic','SQ_IPC','SQ_Long','SQ_Short') fit_IPC_bool <- FALSE } else { colnames1 <- c('Well','Sample','Cq_Converted','Cq_Genomic', 'Cq_IPC','Cq_Long','Cq_Short') colnames2 <- c('Well','Sample','Cq_Converted','Cq_Genomic', 'Cq_IPC','Cq_Long','Cq_Short','SQ_Converted','SQ_Genomic','SQ_IPC','SQ_Long','SQ_Short') fit_IPC_bool <- TRUE } df_Cq_avg <- df_Cq_avg %>% group_by(Well) %>% dplyr::summarize(across(everything(), ~.x[!is.na(.x)][1])) %>% relocate(colnames(df_Cq_avg)) df_std_Cq_avg <- subset(df_Cq_avg, grepl('Std',df_Cq_avg$Sample)) df_std_samples <- subset(df_Cq_avg, !grepl('Std',df_Cq_avg$Sample)) Sample_list <- df_std_samples$Sample Well_list <- df_std_samples$Well df_std_samples <- round(data.frame(lapply(df_std_samples,as.numeric)),2) df_std_samples$Sample <- Sample_list df_std_samples$Well <- Well_list colnames(df_std_samples) <- colnames1 #Add Copy numbers to Cq data frames standards_df <- Create_3x_standards_df(standards_list) df_std_Cq_avg <- subset(df_std_Cq_avg,!grepl('F',df_std_Cq_avg$Sample) & !grepl('G',df_std_Cq_avg$Sample) & !grepl('H',df_std_Cq_avg$Sample)) #only keep top 5 standards for (i in c(1,2,3,4,5)){ standard <- distinct(data.frame(df_std_Cq_avg$Sample))[i,1] df_std_Cq_avg$SQ_Converted[df_std_Cq_avg$Sample==standard] <-t(standards_df[c(3),])[i] df_std_Cq_avg$SQ_Genomic[df_std_Cq_avg$Sample==standard] <-t(standards_df[c(4),])[i] df_std_Cq_avg$SQ_IPC[df_std_Cq_avg$Sample==standard] <-t(standards_df[c(5),])[i] df_std_Cq_avg$SQ_Long[df_std_Cq_avg$Sample==standard] <-t(standards_df[c(1),])[i] df_std_Cq_avg$SQ_Short[df_std_Cq_avg$Sample==standard] <-t(standards_df[c(2),])[i] } colnames(df_std_Cq_avg) <- colnames2 Sample_list <- df_std_Cq_avg$Sample Well_list <- df_std_Cq_avg$Well df_std_Cq_avg <- round(data.frame(lapply(df_std_Cq_avg,as.numeric)),2) df_std_Cq_avg$Sample <- Sample_list df_std_Cq_avg$Well <- Well_list #Automatically exlude a well from the standards which is an outlier based on standard deviation. #subset only wells with standards df_std_Cq <- df_std_Cq_avg #Create standard curves. #Average the Cq values per standard, calculate the log10 values and perform linear regression. df_std_Cq_avg_Converted <- df_std_Cq %>% group_by(Sample) %>% dplyr::summarise(Avg = mean(head(Cq_Converted), na.rm = TRUE)) colnames(df_std_Cq_avg_Converted) <- c('Std','Cq_Converted') df_std_Cq_avg_Genomic <- df_std_Cq %>% group_by(Sample) %>% dplyr::summarise(Avg = mean(head(Cq_Genomic), na.rm = TRUE)) colnames(df_std_Cq_avg_Genomic) <- c('Std','Cq_Genomic') df_std_Cq_avg_IPC <- df_std_Cq %>% group_by(Sample) %>% dplyr::summarise(Avg = mean(head(Cq_IPC), na.rm = TRUE)) colnames(df_std_Cq_avg_IPC) <- c('Std','Cq_IPC') df_std_Cq_avg_Long <- df_std_Cq %>% group_by(Sample) %>% dplyr::summarise(Avg = mean(head(Cq_Long), na.rm = TRUE)) colnames(df_std_Cq_avg_Long) <- c('Std','Cq_Long') df_std_Cq_avg_Short <- df_std_Cq %>% group_by(Sample) %>% dplyr::summarise(Avg = mean(head(Cq_Short), na.rm = TRUE)) colnames(df_std_Cq_avg_Short) <- c('Std','Cq_Short') df_std_Cq_avg <- merge(merge(merge(merge(df_std_Cq_avg_Converted,df_std_Cq_avg_Genomic),df_std_Cq_avg_IPC),df_std_Cq_avg_Long),df_std_Cq_avg_Short) amount_of_standards <- 5 new_col <- t(log10(standards_df[rownames(standards_df)[3],][1:amount_of_standards])) colnames(new_col) = 'SQ_Converted_log' df_std_Cq_avg$SQ_Converted_log <- new_col new_col <- t(log10(standards_df[rownames(standards_df)[4],][1:amount_of_standards])) colnames(new_col) = 'SQ_Genomic_log' df_std_Cq_avg$SQ_Genomic_log <- new_col new_col <-t(log10(standards_df[rownames(standards_df)[5],][1:amount_of_standards])) colnames(new_col) <- 'SQ_IPC_log' df_std_Cq_avg$SQ_IPC_log <- new_col new_col <- t(log10(standards_df[rownames(standards_df)[1],][1:amount_of_standards])) colnames(new_col) <- 'SQ_Long_log' df_std_Cq_avg$SQ_Long_log <- new_col new_col <- t(log10(standards_df[rownames(standards_df)[2],][1:amount_of_standards])) colnames(new_col) <- 'SQ_Short_log' df_std_Cq_avg$SQ_Short_log <- new_col df_std_Cq$SQ_Converted_log <- log10(df_std_Cq$SQ_Converted) df_std_Cq$SQ_Genomic_log <- log10(df_std_Cq$SQ_Genomic) df_std_Cq$SQ_IPC_log <- log10(df_std_Cq$SQ_IPC) df_std_Cq$SQ_Long_log <- log10(df_std_Cq$SQ_Long) df_std_Cq$SQ_Short_log <- log10(df_std_Cq$SQ_Short) Converted.lm <- lm(Cq_Converted ~ SQ_Converted_log, data = df_std_Cq_avg) Genomic.lm <- lm(Cq_Genomic ~ SQ_Genomic_log, data = df_std_Cq_avg) if (fit_IPC_bool){ IPC.lm <- lm(Cq_IPC ~ SQ_IPC_log, data = df_std_Cq_avg) } Long.lm <- lm(Cq_Short ~ SQ_Long_log, data = df_std_Cq_avg) Short.lm <- lm(Cq_Short ~ SQ_Short_log, data = df_std_Cq_avg) Converted_efficiency <- (-1 + 10^(-1/Converted.lm$coefficients[2]))*100 Genomic_efficiency <- (-1 + 10^(-1/Genomic.lm$coefficients[2]))*100 if (fit_IPC_bool){ IPC_efficiency <- (-1 + 10^(-1/IPC.lm$coefficients[2]))*100} Long_efficiency <- (-1 + 10^(-1/Long.lm$coefficients[2]))*100 Short_efficiency <- (-1 + 10^(-1/Short.lm$coefficients[2]))*100 if (fit_IPC_bool){ intercepts <- c(Converted.lm$coefficients[1],Genomic.lm$coefficients[1],IPC.lm$coefficients[1],Long.lm$coefficients[1],Short.lm$coefficients[1]) slopes <- c(Converted.lm$coefficients[2],Genomic.lm$coefficients[2],IPC.lm$coefficients[2],Long.lm$coefficients[2],Short.lm$coefficients[2]) df_std <- data.frame(c(Converted_efficiency,Genomic_efficiency,IPC_efficiency,Long_efficiency,Short_efficiency),c(summary(Converted.lm)$r.squared,summary(Genomic.lm)$r.squared,summary(IPC.lm)$r.squared,summary(Long.lm)$r.squared,summary(Short.lm)$r.squared),intercepts,slopes) colnames(df_std) <- c('PCR efficiency (%)','r_squared','intercept','slope')} else { intercepts <- c(Converted.lm$coefficients[1],Genomic.lm$coefficients[1],Long.lm$coefficients[1],Short.lm$coefficients[1]) slopes <- c(Converted.lm$coefficients[2],Genomic.lm$coefficients[2],Long.lm$coefficients[2],Short.lm$coefficients[2]) df_std <- data.frame(c(Converted_efficiency,Genomic_efficiency,Long_efficiency,Short_efficiency),c(summary(Converted.lm)$r.squared,summary(Genomic.lm)$r.squared,summary(Long.lm)$r.squared,summary(Short.lm)$r.squared),intercepts,slopes) } if (fit_IPC_bool){ df_std_samples$SQ_IPC <- 10^((df_std_samples$Cq_IPC - IPC.lm$coefficients[1])/IPC.lm$coefficients[2]) } df_std_samples$SQ_Genomic <- 10^((df_std_samples$Cq_Genomic - Genomic.lm$coefficients[1])/Genomic.lm$coefficients[2]) df_std_samples$SQ_Long <- 10^((df_std_samples$Cq_Long - Long.lm$coefficients[1])/Long.lm$coefficients[2]) df_std_samples$SQ_Short <- 10^((df_std_samples$Cq_Short - Short.lm$coefficients[1])/Short.lm$coefficients[2]) df_std_samples$SQ_Converted <- 10^((df_std_samples$Cq_Converted - Converted.lm$coefficients[1])/Converted.lm$coefficients[2]) #Calculate the Standard deviation per Sample per fluorophore for (fluor in c('Converted','Genomic','Long','Short')){ Cq_fluor = paste('Cq_',fluor,sep = '') Sd_fluor = paste('SD_',fluor,sep = '') for (level in levels(factor(df_std_samples$Sample))) { level_ind = df_std_samples$Sample==level standard_dev <- sd(df_std_samples[[Cq_fluor]][df_std_samples$Sample==level][!is.na(df_std_samples[[Cq_fluor]][df_std_samples$Sample==level])]) df_std_samples[[Sd_fluor]][df_std_samples$Sample==level] <- standard_dev #When calculating the standard deviation, only the values that are not NA are taken into account. #This means however that the standard deviation will be NA when there is either only one Cq-value left (because there is no standard deviation when you only have one value) or all Cq-values are NA. }} return(p=df_std_samples) } #Exclude sample wells based on the standard deviation by a range around the mean and a standard deviation threshold. exclude_on_SD <- function(df_std_samples,sd_threshold,sd_range){ #First exclude the assay of the samples with a Standard deviation above the set sd_threshold for (fluor in c('Converted','Genomic','Long','Short')){ Cq_fluor = paste('Cq_',fluor,sep = '') Sd_fluor = paste('SD_',fluor,sep = '') Sq_fluor = paste('SQ_',fluor,sep = '') df_std_samples[[Cq_fluor]][df_std_samples[[Sd_fluor]]>sd_threshold] <- NA #set Cq to NA to exclude for qBiCo df_std_samples[[Sq_fluor]][df_std_samples[[Sd_fluor]]>sd_threshold] <- NA #set copy number to NA to exclude for qBiCo } #Next exclude the assay of the wells with a Cq value which is outside the range of sd_range (e.g. 2) times the SD from the average. for (Sample in levels(factor(df_std_samples$Sample))){ for (fluor in c('Converted','Genomic','Long','Short')){ Cq_fluor = paste('Cq_',fluor,sep = '') Sd_fluor = paste('SD_',fluor,sep = '') Sq_fluor = paste('SQ_',fluor,sep = '') Cq_values = df_std_samples[[Cq_fluor]][df_std_samples$Sample==Sample] Mean = mean(Cq_values) Standard_deviation = df_std_samples[[Sd_fluor]][df_std_samples$Sample==Sample][1] #Set Cq-values which are above the range to NA df_std_samples[[Cq_fluor]][df_std_samples$Sample==Sample][Cq_values > (Mean + sd_range*Standard_deviation)] <- NA df_std_samples[[Sq_fluor]][df_std_samples$Sample==Sample][Cq_values > (Mean + sd_range*Standard_deviation)] <- NA #Set Cq-values which are below the range to NA df_std_samples[[Cq_fluor]][df_std_samples$Sample==Sample][Cq_values < (Mean - sd_range*Standard_deviation)] <- NA df_std_samples[[Sq_fluor]][df_std_samples$Sample==Sample][Cq_values < (Mean - sd_range*Standard_deviation)] <- NA }} return(p=df_std_samples) } #Compute the qBiCo indices from the obtained copy numbers per well. parameters_to_indices <- function(raw_data, df_std_samples,experiment_label){ ## CALCULATE QTAP indices from copy numbers ## #Add amounts labels <- subset(raw_data,select = c(Well,DNA_input_ng)) Well_list <- labels$Well labels <- data.frame(lapply(labels,as.numeric)) labels$Well <- Well_list df_std_samples$DNA_input_ng <- with(labels, DNA_input_ng[match(df_std_samples$Well, labels$Well)]) #Add volumes labels <- subset(raw_data,select = c(Well, Elution_volume_ul)) Well_list <- labels$Well labels <- data.frame(lapply(labels,as.numeric)) labels$Well <- Well_list df_std_samples$Elution_volume_ul <- with(labels, Elution_volume_ul[match(df_std_samples$Well, labels$Well)]) #Average the copy numbers for duplicate or triplicate wells. #df_std_samples <- data.frame(df_std_samples %>% group_by(Sample) %>% summarise(mean(SQ_IPC),mean(SQ_Genomic),mean(SQ_Long),mean(SQ_Short),mean(SQ_Converted),mean(DNA_input_ng), mean(Elution_volume_ul))) #colnames(df_std_samples) <- c('Sample', 'SQ_IPC','SQ_Genomic','SQ_Long','SQ_Short','SQ_Converted','DNA_input_ng','Elution_volume_ul') #Short: Short #Long: Long #Converted: Converted #Genomic: Genomic #IPC: Genomic #Calculate conversion efficiency: Converted/(Converted+Genomic) --> Converted*100/(Converted+Genomic/2) df_std_samples$Conversion_efficiency <- df_std_samples$SQ_Converted *100 / (df_std_samples$SQ_Converted + df_std_samples$SQ_Genomic/2) #divide the amount of genomic copies by two as the assay targets both strands of the genomic target, but only 1 of the others. #Calculate recovery: Short / input --> Short / input Avogadro = 6.022E+23 #print('check the size') genome_length = 3200000000 #bp df_std_samples$DNA_conc <- ((df_std_samples$SQ_Short) * genome_length * 10^9 * 617.96 ) / Avogadro #average mass of 1 bp dsDNA is 617.96 g/mole #Calculate fragmentation: (Long / Short)*100 --> (1- Long / Short)*100 df_std_samples$Fragmentation <- (df_std_samples$SQ_Short/ df_std_samples$SQ_Long) #df_qBiCo_indices <- data.frame(df_std_samples %>% group_by(Sample) %>% summarise(mean(Conversion_efficiency,na.rm=TRUE), sd(Conversion_efficiency,na.rm=TRUE), mean(DNA_conc,na.rm=TRUE),sd(DNA_conc,na.rm=TRUE),mean(Fragmentation,na.rm=TRUE),sd(Fragmentation,na.rm=TRUE),mean(DNA_input_ng,na.rm=TRUE), mean(Elution_volume_ul,na.rm=TRUE))) #colnames(df_qBiCo_indices) <- c('Sample','Conversion_efficiency (%)','SD_conversion_efficiency','DNA concentration (ng/µl)','SD_concentration','Fragmentation (%)','SD_fragmentation','Amount (ng)','Elution volume (µl)') df_std_samples$experiment <- experiment_label df_qBiCo_indices <- subset(df_std_samples,select=c(Sample,Well,Conversion_efficiency,DNA_conc,Fragmentation,experiment,DNA_input_ng,Elution_volume_ul)) return(p=df_qBiCo_indices) } #All in one function raw_to_indices <- function(raw_data, experiment_label, sd_threshold, sd_range, manual_exlusion=FALSE,standards_list){ df_std_samples <- raw_to_parameters(raw_data,manual_exlusion,standards_list) #compute the copy numbers and Cq standard deviation. df_std_samples_excluded <- exclude_on_SD(df_std_samples,sd_threshold,sd_range) #exclude samples based on standard deviation. df_qBiCo_indices <- parameters_to_indices(raw_data,df_std_samples_excluded,experiment_label) #obtain qBiCo indices per well. return (p=df_qBiCo_indices) } #Average the qBiCo indices for all wells per sample. Average_indices <- function(df_qBiCo_indices){ df_qBiCo_indices <- data.frame(df_qBiCo_indices %>% group_by(Sample,experiment) %>% dplyr::summarise(mean(Conversion_efficiency,na.rm=TRUE), sd(Conversion_efficiency,na.rm=TRUE), mean(DNA_conc,na.rm=TRUE),sd(DNA_conc,na.rm=TRUE),mean(Fragmentation,na.rm=TRUE),sd(Fragmentation,na.rm=TRUE),mean(DNA_input_ng,na.rm=TRUE), mean(Elution_volume_ul,na.rm=TRUE), )) colnames(df_qBiCo_indices) <- c('Sample','experiment','Conversion_efficiency (%)','SD_conversion_efficiency','DNA concentration (ng/µl)','SD_concentration','Fragmentation','SD_fragmentation','Amount (ng)','Elution volume (µl)') #When only one of the replicates is not exluded, also exclude this replicate as we need 2 for qPCR results. This could happen when for 1 fo the 3 replicates the Cq value for converted assay is excluded and for another replicate the Cq value for the genomic assay. When those are excluded we can only calculate the conversion efficiency for 1 of the 3 replicates. df_qBiCo_indices$Conversion_efficiency[is.na(df_qBiCo_indices$SD_conversion_efficiency)] <- NA df_qBiCo_indices$`DNA concentration (ng/µl)`[is.na(df_qBiCo_indices$SD_concentration)] <- NA df_qBiCo_indices$Fragmentation[is.na(df_qBiCo_indices$SD_fragmentation)] <- NA return(p=df_qBiCo_indices) } Create_3x_standards_df = function(x){ standards_df <- data.frame(x) colnames(standards_df) <- 'StdA' rownames(standards_df) <- c('hTERT_long','hTERT_short','LINE1_converted','LINE1_genomic','IPC') standards_df$StdB <- as.integer(round(standards_df$StdA/3)) standards_df$StdC <- as.integer(round(standards_df$StdB/3)) standards_df$StdD <- as.integer(round(standards_df$StdC/3)) standards_df$StdE <- as.integer(round(standards_df$StdD/3)) standards_df$StdF <- as.integer(round(standards_df$StdE/3)) standards_df$StdG <- as.integer(round(standards_df$StdF/3)) standards_df$StdH <- as.integer(round(standards_df$StdG/3)) return(p = standards_df) } #Double colours double_colour_list <- function(colour_list){ double_colour_list = c() for (colour_code in colour_list){ double_colour_list <- c(double_colour_list,colour_code) double_colour_list <- c(double_colour_list,colour_code) } return(p=double_colour_list) } data_summary <- function(data, varname, groupnames){ require(plyr) summary_func <- function(x, col){ c(mean = mean(x[[col]], na.rm=TRUE), sd = sd(x[[col]], na.rm=TRUE)) } data_sum<-ddply(data, groupnames, .fun=summary_func, varname) data_sum <- plyr::rename(data_sum, c("mean" = varname)) return(data_sum) } #Import CQ data#### ## SET THESE INPUT PARAMETERS ## foldername <- "/media/nw_disk1/roy/comparative_study/data" filename <- paste0(foldername, '/20230206_Comparative_study_Cq_values.xlsx') data <- import_list(filename) BC_1 <- data.frame(data$`BC_1`) BC_2 <- data.frame(data$`BC_2`) BC_3 <- data.frame(data$`BC_3`) BC_4 <- data.frame(data$`BC_4`) BC_5 <- data.frame(data$`BC_5`) BC_6 <- data.frame(data$`BC_6`) BC_7 <- data.frame(data$`BC_7`) EC_1 <- data.frame(data$`EC_1`) EC_2 <- data.frame(data$`EC_2`) EC_3 <- data.frame(data$`EC_3`) EC_5 <- data.frame(data$`EC_5`) EC_6 <- data.frame(data$`EC_6`) EC_7 <- data.frame(data$`EC_7`) EC_8 <- data.frame(data$`EC_8`) BCEC_set1 <- data.frame(data$`gDNA_set1`) BCEC_set2 <- data.frame(data$`gDNA_set2`) graph_info <- data.frame(data$graphs_sample_list) #Get qBiCo indices#### #separately #df_qBiCo_CFX_96_1 <- raw_to_indices(CFX96_1,'CFX96_1',sd_threshold = 1, sd_range = 2) #all experiments in one go sd_threshold = 1000000 sd_range = 2 for (experiment_label in c('BC_1','BC_2','BC_3','BC_4','BC_5','BC_6','BC_7','EC_1','EC_2','EC_3','EC_5','EC_6','EC_7','EC_8','BCEC_set1','BCEC_set2')){ raw_data = get(experiment_label) raw_data <- subset(raw_data ,!is.na(raw_data$Sample)) raw_data <- subset(raw_data,!grepl(1,raw_data$excluded) | grepl('Std',raw_data$Sample)) #raw_data <- subset(raw_data,!raw_data$excluded == 1) output_variable_name <- paste('df_qBiCo_',experiment_label,sep = '') averaged_output_variable_name <- paste('df_averaged_qBiCo_',experiment_label,sep = '') if (experiment_label %in% c('BC_1','EC_1','EC_2','EC_3')){ standards_list <- c(6000,6000,384000,11400,3000) } else {standards_list <- c(6000,12000,768000,22800,3000)} output <- raw_to_indices(raw_data,experiment_label, sd_threshold = sd_threshold, sd_range = sd_range, manual_exlusion=FALSE,standards_list = standards_list) output <- subset(output,output$Sample %in% raw_data$Sample[raw_data$excluded==0]) averaged_output <- Average_indices(output) #average qBiCo indices per sample. assign(output_variable_name, output) assign(averaged_output_variable_name,averaged_output) } # group Sample data df_qBiCo_validation_averaged <- rbind(df_averaged_qBiCo_BC_1,df_averaged_qBiCo_BC_2,df_averaged_qBiCo_BC_3,df_averaged_qBiCo_BC_4,df_averaged_qBiCo_BC_5,df_averaged_qBiCo_BC_6,df_averaged_qBiCo_BC_7,df_averaged_qBiCo_EC_1,df_averaged_qBiCo_EC_2,df_averaged_qBiCo_EC_3,df_averaged_qBiCo_EC_5,df_averaged_qBiCo_EC_6,df_averaged_qBiCo_EC_7,df_averaged_qBiCo_EC_8) df_qBiCo_validation_averaged$Expected_concentration <- df_qBiCo_validation_averaged$`Amount (ng)`/df_qBiCo_validation_averaged$`Elution volume (µl)` # group well data df_qBiCo_validation <- rbind(df_qBiCo_BC_1,df_qBiCo_BC_2,df_qBiCo_BC_3,df_qBiCo_BC_4,df_qBiCo_BC_5,df_qBiCo_BC_6,df_qBiCo_BC_7,df_qBiCo_EC_1,df_qBiCo_EC_2,df_qBiCo_EC_3,df_qBiCo_EC_5,df_qBiCo_EC_6,df_qBiCo_EC_7,df_qBiCo_EC_8) df_qBiCo_validation$Expected_concentration <- df_qBiCo_validation$DNA_input_ng/df_qBiCo_validation$Elution_volume_ul #add conversion_method label df_qBiCo_validation_averaged$conversion_method[df_qBiCo_validation_averaged$experiment %in% c('BC_1','BC_2','BC_3','BC_4','BC_5','BC_6','BC_7','BC_8')] <- 'BC' df_qBiCo_validation_averaged$conversion_method[df_qBiCo_validation_averaged$experiment %in% c('EC_1','EC_2','EC_3','EC_4','EC_5','EC_6','EC_7','EC_8')] <- 'EC' df_qBiCo_validation$conversion_method[df_qBiCo_validation$experiment %in% c('BC_1','BC_2','BC_3','BC_4','BC_5','BC_6','BC_7','BC_8')] <- 'BC' df_qBiCo_validation$conversion_method[df_qBiCo_validation$experiment %in% c('EC_1','EC_2','EC_3','EC_4','EC_5','EC_6','EC_7','EC_8')] <- 'EC' df_qBiCo_validation$Recovery <- (df_qBiCo_validation$DNA_conc*df_qBiCo_validation$Elution_volume_ul)/df_qBiCo_validation$DNA_input_ng #df_qBiCo_validation_averaged$Recovery <- (df_qBiCo_validation_averaged$DNA_conc*df_qBiCo_validation_averaged$Elution_volume_ul)/df_qBiCo_validation_averaged$DNA_input_ng*100 #df_qBiCo_validation$Recovery[df_qBiCo_validation$Recovery > 100] <- 100 #group and label BCEC_set1 and BCEC_set2 data (22 10 ng gDNA conversions) df_qbico_22_averaged <- rbind(df_averaged_qBiCo_BCEC_set1,df_averaged_qBiCo_BCEC_set2) df_qbico_22_averaged$Expected_concentration <- df_qbico_22_averaged$`Amount (ng)`/df_qbico_22_averaged$`Elution volume (µl)` df_qbico_22 <- rbind(df_qBiCo_BCEC_set1,df_qBiCo_BCEC_set2) df_qbico_22$Expected_concentration <- df_qbico_22$DNA_input_ng/df_qbico_22$Elution_volume_ul df_qbico_22_averaged$conversion_method[df_qbico_22_averaged$Sample %in% c('BC','BC_neg')] <- 'BC' df_qbico_22_averaged$conversion_method[df_qbico_22_averaged$Sample %in% c('EC','EC_neg')] <- 'EC' df_qbico_22$conversion_method[df_qbico_22$Sample %in% c('BC','BC_neg')] <- 'BC' df_qbico_22$conversion_method[df_qbico_22$Sample %in% c('EC','EC_neg')] <- 'EC' df_qbico_22$Recovery <- (df_qbico_22$DNA_conc*df_qbico_22$Elution_volume_ul)/df_qbico_22$DNA_input_ng #Plotting & Statistical analysis#### #Create plots for all parameters per index#### #c("repeatability", "sensitivity" , "reproducibility" ,"recovery", "stability_UV", "stability_sonication", "stability_storage", "linearity", "incubation_time","inhibition_hematin","inhibition_proteinase, "freezing"") print_stats <- FALSE print_stats_AB <- TRUE perform_stats <- TRUE point_size <- 2 hide_ns <- FALSE #colours colour_EC <- '#F79B80' colour_BC <- '#00BFC4' colours_conversion_method <- c(colour_BC,colour_EC) #Conversion efficiency#### y_lim_conv = c(45,102) y_lim_coord = c(50,60,70,80,90,100) y_lim_conv_s = c(88,100.5) y_lim_coord_s = c(90,95,100) y_lim_conv_lin = c(0,100) y_lim_coord_lin = c(0,25,50,75,100) x_lim_coord_lin = c(0,25,50,75,100) graph_info_selected <- subset(graph_info,graph_info$index=='conv_eff') #for (instrument_name in subset(graph_info_selected$instrument,!duplicated(graph_info_selected$instrument))) {#create one grob per machine for (parameter_name in c("repeatability" , "reproducibility" ,"recovery", "stability_UV", "stability_sonication", "stability_storage", "linearity", "incubation_time","inhibition_hematin","inhibition_proteinase", "freezing")) {#create one plot per parameter graph_data <- subset(graph_info_selected,graph_info_selected$parameter == parameter_name) x_label = graph_data$x_axis[1] y_label = graph_data$y_axis[1] #select samples/wells #obtain qBiCo indices per well qBiCo_data = df_qBiCo_validation #subset to select only wells that are needed for the plot well_list = graph_data$well run_list = graph_data$run combined_list = paste0(well_list,run_list) qBiCo_data$combined <- paste0(qBiCo_data$Well,qBiCo_data$experiment) qBiCo_data = subset(qBiCo_data,qBiCo_data$combined %in% combined_list) if (parameter_name == 'repeatability') { if (dim(qBiCo_data)[1] == 0){f1 <- ggplot()} #return empty plot if dataframe is empty else{ number_levels1 <- subset(qBiCo_data$DNA_input_ng,!duplicated(qBiCo_data$DNA_input_ng)) #pick amounts for x-axis f1 <- ggplot(qBiCo_data,aes(x=factor(DNA_input_ng, level = number_levels1), y=Conversion_efficiency, color = conversion_method)) f1 <- f1 + geom_boxplot(data=qBiCo_data, outlier.shape = NA,aes(group = interaction(conversion_method,DNA_input_ng)), color = c("#00BFC4","#F79B80","#00BFC4","#F79B80","#00BFC4","#F79B80","#00BFC4","#F79B80","#00BFC4","#F79B80")) f1 <- f1 + scale_color_manual(values = c("#00BFC4","#F79B80","#000000","#000000")) f1 <- f1 + geom_miss_point(aes(shape = conversion_method,group = conversion_method,colour = interaction(..group..,..missing..)),size = point_size/2,position = position_jitterdodge()) f1 <- f1 + coord_cartesian(ylim=y_lim_conv) + scale_y_continuous(breaks=y_lim_coord) f1 <- f1 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f1 <- f1 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f1 <- f1 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) f1 <- f1 + theme(plot.title = element_text(size = 12)) f1 <- f1 + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Conversion_efficiency ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_conv <- append(stat_list_conv, list(stat_list)) q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$DNA_input_ng,q_BC$Conversion_efficiency,'BC'),cbind(q_EC$DNA_input_ng,q_EC$Conversion_efficiency,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- anova_test(data = data.bind, formula = y ~ group, covariate = x)#T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p) stat_list_conv = list(stat_list) #pairwise wilcoxon ranksum test (separate per conversion method) f1 <- f1 + stat_compare_means(method = 'wilcox.test', paired = FALSE, label = 'p.signif', hide.ns = TRUE) stat_list_pairwise <- compare_means(Conversion_efficiency ~ conversion_method, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'DNA_input_ng') stat_list_pairwise_conv = list(stat_list_pairwise) } if (print_stats_AB == TRUE){ p_val <- paste0('p: ',t1$p) f1 <- f1 + annotate('text',x=0.5, y =y_lim_conv[1], label = p_val, vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'reproducibility'){ if (dim(qBiCo_data)[1] == 0) {f2 <- ggplot()} #return empty plot if dataframe is empty else { number_levels2 <- subset(qBiCo_data$DNA_input_ng,!duplicated(qBiCo_data$DNA_input_ng)) #pick amounts for x-axis number_levels2 <- number_levels2[rev(order(number_levels2))] #number_levels2 <- number_levels2[1:minimum_signficicant_conc_index] #only use significantly non different concentrations to compare the 2 runs. qBiCo_data <- subset(qBiCo_data,qBiCo_data$DNA_input_ng %in% number_levels2) qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Conversion_efficiency', groupnames = c('DNA_input_ng','conversion_method','experiment')) f2 <- ggplot(qBiCo_data_plot,aes(x=factor(DNA_input_ng, level = number_levels2), y=Conversion_efficiency, colour = conversion_method, group = experiment)) #f2 <- f2 + facet_grid(cols = vars(factor(DNA_input_ng,level = number_levels2)), scales = 'free', switch = "both") + theme(axis.ticks.x = element_blank(), axis.text.x = element_blank()) f2 <- f2 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f2 <- f2 + coord_cartesian(ylim=y_lim_conv) + scale_y_continuous(breaks=y_lim_coord) f2 <- f2 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f2 <- f2 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold'), legend.title=element_blank()) f2 <- f2 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) legends <- get_legend(f2) f_legend <- as_ggplot(legends) f2 <- f2 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) f2 <- f2 + geom_errorbar(aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd), position = position_dodge(width = 0.9)) #f2 <- f2 + theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) experiments <- subset(qBiCo_data$experiment,!duplicated(qBiCo_data$experiment)) #qBiCo_data_1 <- subset(qBiCo_data,qBiCo_data$experiment == experiments[1]) #f2 <- f2 + stat_compare_means(data = qBiCo_data_1, label.y = 101, label = 'p.signif', method = 't.test', paired = FALSE, ref.group = '5', hide.ns = hide_ns) #qBiCo_data_2 <- subset(qBiCo_data,qBiCo_data$experiment == experiments[2]) #f2 <- f2 + stat_compare_means(data = qBiCo_data_2, label.y = 100, label = 'p.signif', method = 't.test', paired = FALSE, ref.group = '5', hide.ns = hide_ns) #f2 <- f2 + stat_compare_means(data = qBiCo_data_1, method = 'kruskal.test', label.y = 92, label.x = 2) #f2 <- f2 + stat_compare_means(data = qBiCo_data_2, method = 'kruskal.test', label.y = 90, label.x = 2) f2 <- f2 + theme(plot.title = element_text(size = 12)) #We want to compare the two runs if (perform_stats == TRUE){ f2 <- f2 + stat_compare_means(method = 'wilcox.test', paired = FALSE, label = 'p.signif', aes(group = conversion_method), hide.ns = TRUE) #stat_list <- compare_means(Conversion_efficiency ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_conv <- append(stat_list_conv, list(stat_list)) q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$DNA_input_ng,q_BC$Conversion_efficiency,'BC'),cbind(q_EC$DNA_input_ng,q_EC$Conversion_efficiency,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- anova_test(data = data.bind, formula = y ~ group, covariate = x)#T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p) stat_list_conv <- append(stat_list_conv, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Conversion_efficiency ~ conversion_method, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'DNA_input_ng') stat_list_pairwise_conv = append(stat_list_pairwise_conv, list(stat_list_pairwise)) } if (print_stats_AB == TRUE){ p_val <- paste0('p: ',t1$p) f2 <- f2 + annotate('text',x=0.5, y =y_lim_conv[1], label = p_val, vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'recovery') { if (dim(qBiCo_data)[1] == 0) {f3 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$rec_method[grepl('binding5',qBiCo_data$Sample)] <- '5m' qBiCo_data$rec_method[grepl('binding1',qBiCo_data$Sample)] <- '1m' qBiCo_data$rec_method[grepl('double',qBiCo_data$Sample)] <- '2x' qBiCo_data$rec_method[grepl('SHS',qBiCo_data$Sample)] <- 'C' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Conversion_efficiency', groupnames = c('rec_method','conversion_method')) number_levels3 <- c("C","1m","5m",'2x') f3 <- ggplot(qBiCo_data_plot,aes(x=factor(rec_method, level = number_levels3), y=Conversion_efficiency)) f3 <- f3 + geom_miss_point(aes(shape = conversion_method, colour = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f3 <- f3 + facet_grid(.~ conversion_method, scales = 'free') + theme( strip.background = element_blank(), strip.text.x = element_blank()) f3 <- f3 + coord_cartesian(ylim=y_lim_conv_s) + scale_y_continuous(breaks=y_lim_coord_s) f3 <- f3 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f3 <- f3 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f3 <- f3 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f3 <- f3 + aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd) f3 <- f3 + geom_errorbar(aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd, colour = conversion_method), position = position_dodge(width = 0.9)) f3 <- f3 + theme(plot.title = element_text(size = 12)) f3 <- f3 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ BC_pairwise <- compare_means(Conversion_efficiency ~ rec_method, data = subset(qBiCo_data,qBiCo_data$conversion_method == 'BC'), method = 'kruskal.test') EC_pairwise <- compare_means(Conversion_efficiency ~ rec_method, data = subset(qBiCo_data,qBiCo_data$conversion_method == 'EC'), method = 'kruskal.test') BC_pairwise$conversion_method <- 'BC' EC_pairwise$conversion_method <- 'EC' pairwise_tests <- rbind(BC_pairwise,EC_pairwise) stat_list_conv <- append(stat_list_conv, list(c('Kruskal-Wallis',BC_pairwise$p,'BC'))) stat_list_conv <- append(stat_list_conv, list(c('Kruskal-Wallis',EC_pairwise$p,'EC'))) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Conversion_efficiency ~ rec_method, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = 'C', group.by = 'conversion_method') stat_list_pairwise_conv = append(stat_list_pairwise_conv, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f3 <- f3 + stat_compare_means( data = qBiCo_data, method = 'kruskal.test', hide.ns = hide_ns) data_text <- data.frame(label <- paste('p: ',round(pairwise_tests$p,5)), conversion_method <- c('BC','EC')) colnames(data_text) <- c('label','conversion_method') f3 <- f3 + geom_text(data = data_text,mapping = aes(x=0.5, y =y_lim_conv_s[1],label = label) , vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'stability_UV') { if (dim(qBiCo_data)[1] == 0) {f4 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$UV[grepl('30',qBiCo_data$Sample)] <- '30' qBiCo_data$UV[grepl('60',qBiCo_data$Sample)] <- '60' qBiCo_data$UV[grepl('120',qBiCo_data$Sample)] <- '120' qBiCo_data$UV[grepl('SHS',qBiCo_data$Sample)] <- '0' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Conversion_efficiency', groupnames = c('UV','conversion_method')) number_levels4 <- c("0","30","60","120") f4 <- ggplot(qBiCo_data_plot,aes(x=factor(UV, level = number_levels4), y=Conversion_efficiency, colour = conversion_method)) f4 <- f4 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f4 <- f4 + coord_cartesian(ylim=y_lim_conv) + scale_y_continuous(breaks=y_lim_coord) f4 <- f4 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f4 <- f4 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f4 <- f4 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f4 <- f4 + aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd) f4 <- f4 + geom_errorbar(aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd, colour = conversion_method), position = position_dodge(width = 0.9)) f4 <- f4 + theme(plot.title = element_text(size = 12)) f4 <- f4 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Conversion_efficiency ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'UV') #stat_list_conv <- append(stat_list_conv, list(stat_list)) #ANCOVA # res.aov <- qBiCo_data %>% anova_test(Conversion_efficiency ~ UV + conversion_method) # p_val <- paste0('p: ',res.aov$p[2]) # f4 <- f4 + annotate('text',x=0.5, y =y_lim_conv[1], label = p_val, vjust='bottom', hjust = 'left') #Non parametric ANCOVA q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$UV,q_BC$Conversion_efficiency,'BC'),cbind(q_EC$UV,q_EC$Conversion_efficiency,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) p_val <- paste0('p: ',round(t1$p.value,5)) stat_list <- c('p: ',t1$p.value) stat_list_conv <- append(stat_list_conv, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Conversion_efficiency ~ UV, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '0', group.by = 'conversion_method') stat_list_pairwise_conv = append(stat_list_pairwise_conv, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f5 <- f5 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f4 <- f4 + annotate('text',x=0.5, y =y_lim_conv[1], label = p_val, vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'stability_sonication') { if (dim(qBiCo_data)[1] == 0) {f5 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$SON[grepl('SON150',qBiCo_data$Sample)] <- '150' qBiCo_data$SON[grepl('SON500',qBiCo_data$Sample)] <- '500' qBiCo_data$SON[grepl('SON1000',qBiCo_data$Sample)] <- '1000' qBiCo_data$SON[grepl('SON0',qBiCo_data$Sample)] <- 'C' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Conversion_efficiency', groupnames = c('SON','conversion_method')) number_levels5 <- c("C","1000","500","150") f5 <- ggplot(qBiCo_data_plot,aes(x=factor(SON, level = number_levels5), y=Conversion_efficiency, colour = conversion_method)) f5 <- f5 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f5 <- f5 + coord_cartesian(ylim=y_lim_conv_s) + scale_y_continuous(breaks=y_lim_coord_s) f5 <- f5 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f5 <- f5 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f5 <- f5 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f5 <- f5 + aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd) f5 <- f5 + geom_errorbar(aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd), position = position_dodge(width = 0.9)) f5 <- f5 + theme(plot.title = element_text(size = 12)) f5 <- f5 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Conversion_efficiency ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_conv <- append(stat_list_conv, list(stat_list)) qBiCo_data$SON[qBiCo_data$SON == 'C'] <- '2000' q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$SON,q_BC$Conversion_efficiency,'BC'),cbind(q_EC$SON,q_EC$Conversion_efficiency,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_conv <- append(stat_list_conv, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Conversion_efficiency ~ SON, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '2000', group.by = 'conversion_method') stat_list_pairwise_conv = append(stat_list_pairwise_conv, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f5 <- f5 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f5 <- f5 + annotate('text',x=0.5, y =y_lim_conv_s[1], label = p_val, vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'stability_storage') { if (dim(qBiCo_data)[1] == 0) {f6 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$store[grepl('BC_2',qBiCo_data$experiment)] <- 'No storage' qBiCo_data$store[grepl('BC_7',qBiCo_data$experiment)] <- '4 weeks' qBiCo_data$store[grepl('EC_7',qBiCo_data$experiment)] <- 'No storage' qBiCo_data$store[grepl('EC_5',qBiCo_data$experiment)] <- '4 weeks' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Conversion_efficiency', groupnames = c('store','conversion_method')) number_levels6 <- c("No storage","4 weeks") f6 <- ggplot(qBiCo_data_plot,aes(x=factor(store, level = number_levels6), y=Conversion_efficiency, colour = conversion_method)) f6 <- f6 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f6 <- f6 + coord_cartesian(ylim=y_lim_conv_s) + scale_y_continuous(breaks=y_lim_coord_s) f6 <- f6 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f6 <- f6 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f6 <- f6 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f6 <- f6 + aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd) f6 <- f6 + geom_errorbar(aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd), position = position_dodge(width = 0.9)) f6 <- f6 + theme(plot.title = element_text(size = 12)) f6 <- f6 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Conversion_efficiency ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_conv <- append(stat_list_conv, list(stat_list)) qBiCo_data$store[qBiCo_data$store == 'No storage'] <- '0' qBiCo_data$store[qBiCo_data$store == '4 weeks'] <- '4' q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$store,q_BC$Conversion_efficiency,'BC'),cbind(q_EC$store,q_EC$Conversion_efficiency,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_conv <- append(stat_list_conv, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Conversion_efficiency ~ store, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '0', group.by = 'conversion_method') stat_list_pairwise_conv = append(stat_list_pairwise_conv, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f6 <- f6 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f6 <- f6 + annotate('text',x=0.5, y =y_lim_conv_s[1], label = p_val, vjust='bottom', hjust = 'left')} } } else if (parameter_name == 'linearity') { if (dim(qBiCo_data)[1] == 0) {f7 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$Meth_perc[grepl('M0',qBiCo_data$Sample)] <- 0 qBiCo_data$Meth_perc[grepl('M25',qBiCo_data$Sample)] <- 25 qBiCo_data$Meth_perc[grepl('M50',qBiCo_data$Sample)] <- 50 qBiCo_data$Meth_perc[grepl('M75',qBiCo_data$Sample)] <- 75 qBiCo_data$Meth_perc[grepl('M100',qBiCo_data$Sample)] <- 100 qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Conversion_efficiency', groupnames = c('Meth_perc','conversion_method')) number_levels7 <- c(0,25,50,75,100) f7 <- ggplot(qBiCo_data_plot,aes(x=factor(Meth_perc, level = number_levels7), y=Conversion_efficiency, colour = conversion_method)) f7 <- f7 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f7 <- f7 + coord_cartesian(ylim=y_lim_conv) + scale_y_continuous(breaks=y_lim_coord) f7 <- f7 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f7 <- f7 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f7 <- f7 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f7 <- f7 + aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd) f7 <- f7 + geom_errorbar(aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd), position = position_dodge(width = 0.9)) f7 <- f7 + theme(plot.title = element_text(size = 12)) f7 <- f7 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Conversion_efficiency ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_conv <- append(stat_list_conv, list(stat_list)) q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$Meth_perc,q_BC$Conversion_efficiency,'BC'),cbind(q_EC$Meth_perc,q_EC$Conversion_efficiency,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_conv <- append(stat_list_conv, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Conversion_efficiency ~ Meth_perc, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '100', group.by = 'conversion_method') stat_list_pairwise_conv = append(stat_list_pairwise_conv, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f7 <- f7 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f7 <- f7 + annotate('text',x=0.5, y =y_lim_conv[1], label = p_val, vjust='bottom', hjust = 'left')} } } else if (parameter_name == 'incubation_time') { if (dim(qBiCo_data)[1] == 0) {f8 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$incub_time <- 'C' qBiCo_data$incub_time[grepl('12h',qBiCo_data$Sample)] <- '-' qBiCo_data$incub_time[grepl('20h',qBiCo_data$Sample)] <- '+' qBiCo_data$incub_time[grepl('APOBEC',qBiCo_data$Sample)] <- '2+' qBiCo_data$incub_time[grepl('TET',qBiCo_data$Sample)] <- '1+' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Conversion_efficiency', groupnames = c('incub_time','conversion_method')) number_levels8 <- c("C",'-','+','1+', '2+') f8 <- ggplot(qBiCo_data_plot,aes(x=factor(incub_time, level = number_levels8), y=Conversion_efficiency)) f8 <- f8 + geom_miss_point(aes(shape = conversion_method, colour = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f8 <- f8 + facet_grid(.~ conversion_method, scales = 'free') + theme( strip.background = element_blank(), strip.text.x = element_blank()) f8 <- f8 + coord_cartesian(ylim=y_lim_conv_s) + scale_y_continuous(breaks=y_lim_coord_s) f8 <- f8 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f8 <- f8 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f8 <- f8 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f8 <- f8 + aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd) f8 <- f8 + geom_errorbar(aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd, colour = conversion_method), position = position_dodge(width = 0.9)) f8 <- f8 + theme(plot.title = element_text(size = 12)) f8 <- f8 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ BC_pairwise <- compare_means(Conversion_efficiency ~ incub_time, data = subset(qBiCo_data,qBiCo_data$conversion_method == 'BC'), method = 'kruskal.test') EC_pairwise <- compare_means(Conversion_efficiency ~ incub_time, data = subset(qBiCo_data,qBiCo_data$conversion_method == 'EC'), method = 'kruskal.test') stat_list_conv <- append(stat_list_conv, list(c('Kruskal-Wallis',BC_pairwise$p,'BC'))) stat_list_conv <- append(stat_list_conv, list(c('Kruskal-Wallis',EC_pairwise$p,'EC'))) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Conversion_efficiency ~ incub_time, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = 'C', group.by = 'conversion_method') stat_list_pairwise_conv = append(stat_list_pairwise_conv, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f8 <- f8 + stat_compare_means(aes(colour = experiment),data = subset(qBiCo_data,qBiCo_data$conversion_method == 'BC'), method = 'kruskal.test', hide.ns = hide_ns) #f8 <- f8 + stat_compare_means(aes(colour = experiment),data = subset(qBiCo_data,qBiCo_data$conversion_method == 'EC'), method = 'kruskal.test', hide.ns = hide_ns) data_text <- data.frame(label <- paste('p: ',round(pairwise_tests$p,5)), conversion_method <- c('BC','EC')) colnames(data_text) <- c('label','conversion_method') f8 <- f8 + geom_text(data = data_text,mapping = aes(x=0.5, y =y_lim_conv_s[1],label = label) , vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'inhibition_hematin') { if (dim(qBiCo_data)[1] == 0) {f9 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$hematin[grepl('HH',qBiCo_data$Sample)] <- '200' qBiCo_data$hematin[grepl('MH',qBiCo_data$Sample)] <- '100' qBiCo_data$hematin[grepl('NH',qBiCo_data$Sample)] <- '0' qBiCo_data$hematin[grepl('SHS',qBiCo_data$Sample)] <- 'C' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Conversion_efficiency', groupnames = c('hematin','conversion_method')) number_levels9 <- c("C","0","100","200") f9 <- ggplot(qBiCo_data_plot,aes(x=factor(hematin, level = number_levels9), y=Conversion_efficiency, colour = conversion_method)) f9 <- f9 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f9 <- f9 + coord_cartesian(ylim=y_lim_conv_s) + scale_y_continuous(breaks=y_lim_coord_s) f9 <- f9 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f9 <- f9 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f9 <- f9 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f9 <- f9 + aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd) f9 <- f9 + geom_errorbar(aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd), position = position_dodge(width = 0.9)) f9 <- f9 + theme(plot.title = element_text(size = 12)) f9 <- f9 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Conversion_efficiency ~ hematin, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = 'C', group.by = 'conversion_method') stat_list_pairwise_conv = append(stat_list_pairwise_conv, list(stat_list_pairwise)) #stat_list <- compare_means(Conversion_efficiency ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_conv <- append(stat_list_conv, list(stat_list)) qBiCo_data <- subset(qBiCo_data,!qBiCo_data$hematin == 'C') q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$hematin,q_BC$Conversion_efficiency,'BC'),cbind(q_EC$hematin,q_EC$Conversion_efficiency,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_conv <- append(stat_list_conv, list(stat_list)) } if (print_stats == TRUE){ #f9 <- f9 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f9 <- f9 + annotate('text',x=0.5, y =y_lim_conv_s[1], label = p_val, vjust='bottom', hjust = 'left')} } } else if (parameter_name == 'inhibition_proteinase') { if (dim(qBiCo_data)[1] == 0) {f10 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$proteinase[grepl('HP',qBiCo_data$Sample)] <- '0.2' qBiCo_data$proteinase[grepl('MP',qBiCo_data$Sample)] <- '0.1' qBiCo_data$proteinase[grepl('LP',qBiCo_data$Sample)] <- '0.05' qBiCo_data$proteinase[grepl('SHS',qBiCo_data$Sample)] <- 'C' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Conversion_efficiency', groupnames = c('proteinase','conversion_method')) number_levels10 <- c("C","0.05","0.1","0.2") f10 <- ggplot(qBiCo_data_plot,aes(x=factor(proteinase, level = number_levels10), y=Conversion_efficiency, colour = conversion_method)) f10 <- f10 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f10 <- f10 + coord_cartesian(ylim=y_lim_conv_s) + scale_y_continuous(breaks=y_lim_coord_s) f10 <- f10 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f10 <- f10 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f10 <- f10 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f10 <- f10 + aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd) f10 <- f10 + geom_errorbar(aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd), position = position_dodge(width = 0.9)) f10 <- f10 + theme(plot.title = element_text(size = 12)) f10 <- f10 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Conversion_efficiency ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_conv <- append(stat_list_conv, list(stat_list)) qBiCo_data$proteinase[qBiCo_data$proteinase == 'C'] <- '0' q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$proteinase,q_BC$Conversion_efficiency,'BC'),cbind(q_EC$proteinase,q_EC$Conversion_efficiency,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_conv <- append(stat_list_conv, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Conversion_efficiency ~ proteinase, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '0', group.by = 'conversion_method') stat_list_pairwise_conv = append(stat_list_pairwise_conv, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f10 <- f10 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f10 <- f10 + annotate('text',x=0.5, y =y_lim_conv_s[1], label = p_val, vjust='bottom', hjust = 'left')} } } else if (parameter_name == 'freezing') { if (dim(qBiCo_data)[1] == 0) {f11 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$freezing[grepl('5ft',qBiCo_data$Sample)] <- '5' qBiCo_data$freezing[grepl('10ft',qBiCo_data$Sample)] <- '10' qBiCo_data$freezing[grepl('SHS',qBiCo_data$Sample)] <- '0' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Conversion_efficiency', groupnames = c('freezing','conversion_method')) number_levels11 <- c("0","5","10") f11 <- ggplot(qBiCo_data_plot,aes(x=factor(freezing, level = number_levels11), y=Conversion_efficiency, colour = conversion_method)) f11 <- f11 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f11 <- f11 + coord_cartesian(ylim=y_lim_conv_s) + scale_y_continuous(breaks=y_lim_coord_s) f11 <- f11 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f11 <- f11 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f11 <- f11 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f11 <- f11 + aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd) f11 <- f11 + geom_errorbar(aes(ymax = Conversion_efficiency + sd, ymin = Conversion_efficiency - sd), position = position_dodge(width = 0.9)) f11 <- f11 + theme(plot.title = element_text(size = 12)) f11 <- f11 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Conversion_efficiency ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_conv <- append(stat_list_conv, list(stat_list)) q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$freezing,q_BC$Conversion_efficiency,'BC'),cbind(q_EC$freezing,q_EC$Conversion_efficiency,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_conv <- append(stat_list_conv, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Conversion_efficiency ~ freezing, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '0', group.by = 'conversion_method') stat_list_pairwise_conv = append(stat_list_pairwise_conv, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f11 <- f11 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f11 <- f11 + annotate('text',x=0.5, y =y_lim_conv_s[1], label = p_val, vjust='bottom', hjust = 'left')} } } else {TRUE} } foldername_tiff <- paste0(foldername, '/TIFF') width_tiff <- 3 height_tiff <- 3 ggsave(file = paste0(foldername_tiff,"/Conversion_efficiency/",'repeatability', ".TIFF"), f1, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Conversion_efficiency/",'reproducibility', ".TIFF"), f2, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Conversion_efficiency/",'recovery', ".TIFF"), f3, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Conversion_efficiency/",'stability_UV', ".TIFF"), f4, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Conversion_efficiency/",'stability_sonication', ".TIFF"), f5, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Conversion_efficiency/",'stability_storage', ".TIFF"), f6, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Conversion_efficiency/",'linearity', ".TIFF"), f7, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Conversion_efficiency/",'incubation_time', ".TIFF"), f8, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Conversion_efficiency/",'inhibition_hematin', ".TIFF"), f9, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Conversion_efficiency/",'inhibition_proteinase', ".TIFF"), f10, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Conversion_efficiency/",'freezing', ".TIFF"), f11, width = width_tiff, height = height_tiff) title_g <- 'Conversion efficiency' y_left <- richtext_grob("**Conversion efficiency (%)**", rot=90, gp = gpar(fontsize = 16)) for (figure_lab in c('f1','f2','f3','f8','f7','f4','f5','f9','f10','f6','f11')){ index = which(c('f1','f2','f3','f8','f7','f4','f5','f9','f10','f6','f11')==figure_lab) assign(figure_lab,get(figure_lab) + labs(tag = LETTERS[index])) } g <- grid.arrange(arrangeGrob(f1,f2,f_legend, nrow = 1, widths = c(5,5,1)), arrangeGrob(f3,f8,f7,f4,f5,f9,f10,f6,f11, nrow = 3), nrow = 2, heights = c(2,5), left = y_left) output_g_name <- 'g_conv' assign(output_g_name,g) output_stat_name <- 'stat_conv' assign(output_stat_name,stat_list_conv) #Recovery#### y_lim_rec = c(-0.5,7.5) y_lim_coord = c(0,1,2,3,4,5,6,7) y_lim_rec_s = c(-0.5,5) y_lim_coord_s = c(0,1,2,3,4,5) y_lim_rec_lin = c(-0.5,5) y_lim_coord_lin = c(0,1,2,3,4,5) x_lim_coord_lin = c(0,25,50,75,100) graph_info_selected <- subset(graph_info,graph_info$index=='recovery') #for (instrument_name in subset(graph_info_selected$instrument,!duplicated(graph_info_selected$instrument))) {#create one grob per machine for (parameter_name in c("repeatability" , "reproducibility" ,"recovery", "stability_UV", "stability_sonication", "stability_storage", "linearity", "incubation_time","inhibition_hematin","inhibition_proteinase", "freezing")) {#create one plot per parameter graph_data <- subset(graph_info_selected,graph_info_selected$parameter == parameter_name) x_label = graph_data$x_axis[1] y_label = graph_data$y_axis[1] #select samples/wells #obtain qBiCo indices per well qBiCo_data = df_qBiCo_validation #subset to select only wells that are needed for the plot well_list = graph_data$well run_list = graph_data$run combined_list = paste0(well_list,run_list) qBiCo_data$combined <- paste0(qBiCo_data$Well,qBiCo_data$experiment) qBiCo_data = subset(qBiCo_data,qBiCo_data$combined %in% combined_list) if (parameter_name == 'repeatability') { if (dim(qBiCo_data)[1] == 0){f1 <- ggplot()} #return empty plot if dataframe is empty else{ number_levels1 <- subset(qBiCo_data$DNA_input_ng,!duplicated(qBiCo_data$DNA_input_ng)) #pick amounts for x-axis f1 <- ggplot(qBiCo_data,aes(x=factor(DNA_input_ng, level = number_levels1), y=Recovery, color = conversion_method)) f1 <- f1 + geom_boxplot(data=qBiCo_data, outlier.shape = NA,position = position_dodge(preserve = "single"),aes(group = interaction(conversion_method,DNA_input_ng)), color = c("#00BFC4","#F79B80","#00BFC4","#F79B80","#00BFC4","#F79B80","#00BFC4","#F79B80","#00BFC4","#F79B80")) f1 <- f1 + scale_color_manual(values = c("#000000","#000000","#00BFC4","#F79B80")) f1 <- f1 + geom_miss_point(aes(shape = conversion_method,group = conversion_method,colour = interaction(..group..,..missing..)),size = point_size/2,position = position_jitterdodge()) f1 <- f1 + coord_cartesian(ylim=y_lim_rec) + scale_y_continuous(breaks=y_lim_coord) f1 <- f1 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f1 <- f1 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f1 <- f1 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) f1 <- f1 + theme(plot.title = element_text(size = 12)) f1 <- f1 + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Recovery ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_rec <- append(stat_list_rec, list(stat_list)) q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$DNA_input_ng,q_BC$Recovery,'BC'),cbind(q_EC$DNA_input_ng,q_EC$Recovery,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- anova_test(data = data.bind, formula = y ~ group, covariate = x)#T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p) stat_list_rec = list(stat_list) #pairwise wilcoxon ranksum test (separate per conversion method) f1 <- f1 + stat_compare_means(method = 'wilcox.test', paired = FALSE, label = 'p.signif', hide.ns = TRUE) stat_list_pairwise <- compare_means(Recovery ~ conversion_method, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'DNA_input_ng') stat_list_pairwise_rec = list(stat_list_pairwise) } if (print_stats_AB == TRUE){ p_val <- paste0('p: ',t1$p) f1 <- f1 + annotate('text',x=0.5, y =y_lim_rec[1], label = p_val, vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'reproducibility'){ if (dim(qBiCo_data)[1] == 0) {f2 <- ggplot()} #return empty plot if dataframe is empty else { number_levels2 <- subset(qBiCo_data$DNA_input_ng,!duplicated(qBiCo_data$DNA_input_ng)) #pick amounts for x-axis number_levels2 <- number_levels2[rev(order(number_levels2))] #number_levels2 <- number_levels2[1:minimum_signficicant_conc_index] #only use significantly non different concentrations to compare the 2 runs. qBiCo_data <- subset(qBiCo_data,qBiCo_data$DNA_input_ng %in% number_levels2) qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Recovery', groupnames = c('DNA_input_ng','conversion_method','experiment')) f2 <- ggplot(qBiCo_data_plot,aes(x=factor(DNA_input_ng, level = number_levels2), y=Recovery, colour = conversion_method, group = experiment)) #f2 <- f2 + facet_grid(cols = vars(factor(DNA_input_ng,level = number_levels2)), scales = 'free', switch = "both") + theme(axis.ticks.x = element_blank(), axis.text.x = element_blank()) f2 <- f2 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f2 <- f2 + coord_cartesian(ylim=y_lim_rec) + scale_y_continuous(breaks=y_lim_coord) f2 <- f2 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f2 <- f2 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold'), legend.title=element_blank()) f2 <- f2 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) legends <- get_legend(f2) f_legend <- as_ggplot(legends) f2 <- f2 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) f2 <- f2 + geom_errorbar(aes(ymax = Recovery + sd, ymin = Recovery - sd), position = position_dodge(width = 0.9)) #f2 <- f2 + theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) experiments <- subset(qBiCo_data$experiment,!duplicated(qBiCo_data$experiment)) #qBiCo_data_1 <- subset(qBiCo_data,qBiCo_data$experiment == experiments[1]) #f2 <- f2 + stat_compare_means(data = qBiCo_data_1, label.y = 101, label = 'p.signif', method = 't.test', paired = FALSE, ref.group = '5', hide.ns = hide_ns) #qBiCo_data_2 <- subset(qBiCo_data,qBiCo_data$experiment == experiments[2]) #f2 <- f2 + stat_compare_means(data = qBiCo_data_2, label.y = 100, label = 'p.signif', method = 't.test', paired = FALSE, ref.group = '5', hide.ns = hide_ns) #f2 <- f2 + stat_compare_means(data = qBiCo_data_1, method = 'kruskal.test', label.y = 92, label.x = 2) #f2 <- f2 + stat_compare_means(data = qBiCo_data_2, method = 'kruskal.test', label.y = 90, label.x = 2) f2 <- f2 + theme(plot.title = element_text(size = 12)) #We want to compare the two runs if (perform_stats == TRUE){ f2 <- f2 + stat_compare_means(method = 'wilcox.test', paired = FALSE, label = 'p.signif', aes(group = conversion_method), hide.ns = TRUE) #stat_list <- compare_means(Recovery ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_rec <- append(stat_list_rec, list(stat_list)) q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$DNA_input_ng,q_BC$Recovery,'BC'),cbind(q_EC$DNA_input_ng,q_EC$Recovery,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- anova_test(data = data.bind, formula = y ~ group, covariate = x)#T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p) stat_list_rec <- append(stat_list_rec, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Recovery ~ conversion_method, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'DNA_input_ng') stat_list_pairwise_rec = append(stat_list_pairwise_rec, list(stat_list_pairwise)) } if (print_stats_AB == TRUE){ p_val <- paste0('p: ',t1$p) f2 <- f2 + annotate('text',x=0.5, y =y_lim_rec[1], label = p_val, vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'recovery') { if (dim(qBiCo_data)[1] == 0) {f3 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$rec_method[grepl('binding5',qBiCo_data$Sample)] <- '5m' qBiCo_data$rec_method[grepl('binding1',qBiCo_data$Sample)] <- '1m' qBiCo_data$rec_method[grepl('double',qBiCo_data$Sample)] <- '2x' qBiCo_data$rec_method[grepl('SHS',qBiCo_data$Sample)] <- 'C' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Recovery', groupnames = c('rec_method','conversion_method')) number_levels3 <- c("C","1m","5m",'2x') f3 <- ggplot(qBiCo_data_plot,aes(x=factor(rec_method, level = number_levels3), y=Recovery)) f3 <- f3 + geom_miss_point(aes(shape = conversion_method, colour = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f3 <- f3 + facet_grid(.~ conversion_method, scales = 'free') + theme( strip.background = element_blank(), strip.text.x = element_blank()) f3 <- f3 + coord_cartesian(ylim=y_lim_rec_s) + scale_y_continuous(breaks=y_lim_coord_s) f3 <- f3 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f3 <- f3 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f3 <- f3 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f3 <- f3 + aes(ymax = Recovery + sd, ymin = Recovery - sd) f3 <- f3 + geom_errorbar(aes(ymax = Recovery + sd, ymin = Recovery - sd, colour = conversion_method), position = position_dodge(width = 0.9)) f3 <- f3 + theme(plot.title = element_text(size = 12)) f3 <- f3 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ BC_pairwise <- compare_means(Recovery ~ rec_method, data = subset(qBiCo_data,qBiCo_data$conversion_method == 'BC'), method = 'kruskal.test') EC_pairwise <- compare_means(Recovery ~ rec_method, data = subset(qBiCo_data,qBiCo_data$conversion_method == 'EC'), method = 'kruskal.test') BC_pairwise$conversion_method <- 'BC' EC_pairwise$conversion_method <- 'EC' pairwise_tests <- rbind(BC_pairwise,EC_pairwise) stat_list_rec <- append(stat_list_rec, list(c('Kruskal-Wallis',BC_pairwise$p,'BC'))) stat_list_rec <- append(stat_list_rec, list(c('Kruskal-Wallis',EC_pairwise$p,'EC'))) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Recovery ~ rec_method, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = 'C', group.by = 'conversion_method') stat_list_pairwise_rec = append(stat_list_pairwise_rec, list(stat_list_pairwise))} if (print_stats == TRUE){ #f3 <- f3 + stat_compare_means( data = qBiCo_data, method = 'kruskal.test', hide.ns = hide_ns) data_text <- data.frame(label <- paste('p: ',round(pairwise_tests$p,5)), conversion_method <- c('BC','EC')) colnames(data_text) <- c('label','conversion_method') f3 <- f3 + geom_text(data = data_text,mapping = aes(x=0.5, y =y_lim_rec_s[1],label = label) , vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'stability_UV') { if (dim(qBiCo_data)[1] == 0) {f4 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$UV[grepl('30',qBiCo_data$Sample)] <- '30' qBiCo_data$UV[grepl('60',qBiCo_data$Sample)] <- '60' qBiCo_data$UV[grepl('120',qBiCo_data$Sample)] <- '120' qBiCo_data$UV[grepl('SHS',qBiCo_data$Sample)] <- '0' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Recovery', groupnames = c('UV','conversion_method')) number_levels4 <- c("0","30","60","120") f4 <- ggplot(qBiCo_data_plot,aes(x=factor(UV, level = number_levels4), y=Recovery, colour = conversion_method)) f4 <- f4 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f4 <- f4 + coord_cartesian(ylim=y_lim_rec_s) + scale_y_continuous(breaks=y_lim_coord) f4 <- f4 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f4 <- f4 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f4 <- f4 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f4 <- f4 + aes(ymax = Recovery + sd, ymin = Recovery - sd) f4 <- f4 + geom_errorbar(aes(ymax = Recovery + sd, ymin = Recovery - sd, colour = conversion_method), position = position_dodge(width = 0.9)) f4 <- f4 + theme(plot.title = element_text(size = 12)) f4 <- f4 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Recovery ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'UV') #stat_list_rec <- append(stat_list_rec, list(stat_list)) #ANCOVA # res.aov <- qBiCo_data %>% anova_test(Recovery ~ UV + conversion_method) # p_val <- paste0('p: ',res.aov$p[2]) # f4 <- f4 + annotate('text',x=0.5, y =y_lim_rec[1], label = p_val, vjust='bottom', hjust = 'left') #Non parametric ANCOVA q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$UV,q_BC$Recovery,'BC'),cbind(q_EC$UV,q_EC$Recovery,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) p_val <- paste0('p: ',round(t1$p.value,5)) stat_list <- c('p: ',t1$p.value) stat_list_rec <- append(stat_list_rec, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Recovery ~ UV, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '0', group.by = 'conversion_method') stat_list_pairwise_rec = append(stat_list_pairwise_rec, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f5 <- f5 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f4 <- f4 + annotate('text',x=0.5, y =y_lim_rec[1], label = p_val, vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'stability_sonication') { if (dim(qBiCo_data)[1] == 0) {f5 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$SON[grepl('SON150',qBiCo_data$Sample)] <- '150' qBiCo_data$SON[grepl('SON500',qBiCo_data$Sample)] <- '500' qBiCo_data$SON[grepl('SON1000',qBiCo_data$Sample)] <- '1000' qBiCo_data$SON[grepl('SON0',qBiCo_data$Sample)] <- 'C' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Recovery', groupnames = c('SON','conversion_method')) number_levels5 <- c("C","1000","500","150") f5 <- ggplot(qBiCo_data_plot,aes(x=factor(SON, level = number_levels5), y=Recovery, colour = conversion_method)) f5 <- f5 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f5 <- f5 + coord_cartesian(ylim=y_lim_rec_s) + scale_y_continuous(breaks=y_lim_coord_s) f5 <- f5 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f5 <- f5 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f5 <- f5 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f5 <- f5 + aes(ymax = Recovery + sd, ymin = Recovery - sd) f5 <- f5 + geom_errorbar(aes(ymax = Recovery + sd, ymin = Recovery - sd), position = position_dodge(width = 0.9)) f5 <- f5 + theme(plot.title = element_text(size = 12)) f5 <- f5 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Recovery ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_rec <- append(stat_list_rec, list(stat_list)) qBiCo_data$SON[qBiCo_data$SON == 'C'] <- '2000' q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$SON,q_BC$Recovery,'BC'),cbind(q_EC$SON,q_EC$Recovery,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_rec <- append(stat_list_rec, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Recovery ~ SON, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '2000', group.by = 'conversion_method') stat_list_pairwise_rec = append(stat_list_pairwise_rec, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f5 <- f5 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f5 <- f5 + annotate('text',x=0.5, y =y_lim_rec_s[1], label = p_val, vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'stability_storage') { if (dim(qBiCo_data)[1] == 0) {f6 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$store[grepl('BC_2',qBiCo_data$experiment)] <- 'No storage' qBiCo_data$store[grepl('BC_7',qBiCo_data$experiment)] <- '4 weeks' qBiCo_data$store[grepl('EC_7',qBiCo_data$experiment)] <- 'No storage' qBiCo_data$store[grepl('EC_5',qBiCo_data$experiment)] <- '4 weeks' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Recovery', groupnames = c('store','conversion_method')) number_levels6 <- c("No storage","4 weeks") f6 <- ggplot(qBiCo_data_plot,aes(x=factor(store, level = number_levels6), y=Recovery, colour = conversion_method)) f6 <- f6 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f6 <- f6 + coord_cartesian(ylim=y_lim_rec_s) + scale_y_continuous(breaks=y_lim_coord_s) f6 <- f6 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f6 <- f6 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f6 <- f6 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f6 <- f6 + aes(ymax = Recovery + sd, ymin = Recovery - sd) f6 <- f6 + geom_errorbar(aes(ymax = Recovery + sd, ymin = Recovery - sd), position = position_dodge(width = 0.9)) f6 <- f6 + theme(plot.title = element_text(size = 12)) f6 <- f6 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Recovery ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_rec <- append(stat_list_rec, list(stat_list)) qBiCo_data$store[qBiCo_data$store == 'No storage'] <- '0' qBiCo_data$store[qBiCo_data$store == '4 weeks'] <- '4' q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$store,q_BC$Recovery,'BC'),cbind(q_EC$store,q_EC$Recovery,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_rec <- append(stat_list_rec, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Recovery ~ store, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '0', group.by = 'conversion_method') stat_list_pairwise_rec = append(stat_list_pairwise_rec, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f6 <- f6 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f6 <- f6 + annotate('text',x=0.5, y =y_lim_rec_s[1], label = p_val, vjust='bottom', hjust = 'left')} } } else if (parameter_name == 'linearity') { if (dim(qBiCo_data)[1] == 0) {f7 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$Meth_perc[grepl('M0',qBiCo_data$Sample)] <- 0 qBiCo_data$Meth_perc[grepl('M25',qBiCo_data$Sample)] <- 25 qBiCo_data$Meth_perc[grepl('M50',qBiCo_data$Sample)] <- 50 qBiCo_data$Meth_perc[grepl('M75',qBiCo_data$Sample)] <- 75 qBiCo_data$Meth_perc[grepl('M100',qBiCo_data$Sample)] <- 100 qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Recovery', groupnames = c('Meth_perc','conversion_method')) number_levels7 <- c(0,25,50,75,100) f7 <- ggplot(qBiCo_data_plot,aes(x=factor(Meth_perc, level = number_levels7), y=Recovery, colour = conversion_method)) f7 <- f7 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f7 <- f7 + coord_cartesian(ylim=y_lim_rec_s) + scale_y_continuous(breaks=y_lim_coord) f7 <- f7 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f7 <- f7 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f7 <- f7 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f7 <- f7 + aes(ymax = Recovery + sd, ymin = Recovery - sd) f7 <- f7 + geom_errorbar(aes(ymax = Recovery + sd, ymin = Recovery - sd), position = position_dodge(width = 0.9)) f7 <- f7 + theme(plot.title = element_text(size = 12)) f7 <- f7 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Recovery ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_rec <- append(stat_list_rec, list(stat_list)) q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$Meth_perc,q_BC$Recovery,'BC'),cbind(q_EC$Meth_perc,q_EC$Recovery,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_rec <- append(stat_list_rec, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Recovery ~ Meth_perc, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '100', group.by = 'conversion_method') stat_list_pairwise_rec = append(stat_list_pairwise_rec, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f7 <- f7 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f7 <- f7 + annotate('text',x=0.5, y =y_lim_rec[1], label = p_val, vjust='bottom', hjust = 'left')} } } else if (parameter_name == 'incubation_time') { if (dim(qBiCo_data)[1] == 0) {f8 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$incub_time<- 'C' qBiCo_data$incub_time[grepl('12h',qBiCo_data$Sample)] <- '-' qBiCo_data$incub_time[grepl('20h',qBiCo_data$Sample)] <- '+' qBiCo_data$incub_time[grepl('APOBEC',qBiCo_data$Sample)] <- '2+' qBiCo_data$incub_time[grepl('TET',qBiCo_data$Sample)] <- '1+' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Recovery', groupnames = c('incub_time','conversion_method')) number_levels8 <- c("C",'-','+','1+', '2+') f8 <- ggplot(qBiCo_data_plot,aes(x=factor(incub_time, level = number_levels8), y=Recovery)) f8 <- f8 + geom_miss_point(aes(shape = conversion_method, colour = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f8 <- f8 + facet_grid(.~ conversion_method, scales = 'free') + theme( strip.background = element_blank(), strip.text.x = element_blank()) f8 <- f8 + coord_cartesian(ylim=y_lim_rec_s) + scale_y_continuous(breaks=y_lim_coord_s) f8 <- f8 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f8 <- f8 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f8 <- f8 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f8 <- f8 + aes(ymax = Recovery + sd, ymin = Recovery - sd) f8 <- f8 + geom_errorbar(aes(ymax = Recovery + sd, ymin = Recovery - sd, colour = conversion_method), position = position_dodge(width = 0.9)) f8 <- f8 + theme(plot.title = element_text(size = 12)) f8 <- f8 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ BC_pairwise <- compare_means(Recovery ~ incub_time, data = subset(qBiCo_data,qBiCo_data$conversion_method == 'BC'), method = 'kruskal.test') EC_pairwise <- compare_means(Recovery ~ incub_time, data = subset(qBiCo_data,qBiCo_data$conversion_method == 'EC'), method = 'kruskal.test') stat_list_rec <- append(stat_list_rec, list(c('Kruskal-Wallis',BC_pairwise$p,'BC'))) stat_list_rec <- append(stat_list_rec, list(c('Kruskal-Wallis',EC_pairwise$p,'EC'))) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Recovery ~ incub_time, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = 'C', group.by = 'conversion_method') stat_list_pairwise_rec = append(stat_list_pairwise_rec, list(stat_list_pairwise))} if (print_stats == TRUE){ #f8 <- f8 + stat_compare_means(aes(colour = experiment),data = subset(qBiCo_data,qBiCo_data$conversion_method == 'BC'), method = 'kruskal.test', hide.ns = hide_ns) #f8 <- f8 + stat_compare_means(aes(colour = experiment),data = subset(qBiCo_data,qBiCo_data$conversion_method == 'EC'), method = 'kruskal.test', hide.ns = hide_ns) data_text <- data.frame(label <- paste('p: ',round(pairwise_tests$p,5)), conversion_method <- c('BC','EC')) colnames(data_text) <- c('label','conversion_method') f8 <- f8 + geom_text(data = data_text,mapping = aes(x=0.5, y =y_lim_rec_s[1],label = label) , vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'inhibition_hematin') { if (dim(qBiCo_data)[1] == 0) {f9 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$hematin[grepl('HH',qBiCo_data$Sample)] <- '200' qBiCo_data$hematin[grepl('MH',qBiCo_data$Sample)] <- '100' qBiCo_data$hematin[grepl('NH',qBiCo_data$Sample)] <- '0' qBiCo_data$hematin[grepl('SHS',qBiCo_data$Sample)] <- 'C' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Recovery', groupnames = c('hematin','conversion_method')) number_levels9 <- c("C","0","100","200") f9 <- ggplot(qBiCo_data_plot,aes(x=factor(hematin, level = number_levels9), y=Recovery, colour = conversion_method)) f9 <- f9 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f9 <- f9 + coord_cartesian(ylim=y_lim_rec_s) + scale_y_continuous(breaks=y_lim_coord_s) f9 <- f9 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f9 <- f9 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f9 <- f9 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f9 <- f9 + aes(ymax = Recovery + sd, ymin = Recovery - sd) f9 <- f9 + geom_errorbar(aes(ymax = Recovery + sd, ymin = Recovery - sd), position = position_dodge(width = 0.9)) f9 <- f9 + theme(plot.title = element_text(size = 12)) f9 <- f9 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Recovery ~ hematin, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = 'C', group.by = 'conversion_method') stat_list_pairwise_rec = append(stat_list_pairwise_rec, list(stat_list_pairwise)) #stat_list <- compare_means(Recovery ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_rec <- append(stat_list_rec, list(stat_list)) qBiCo_data <- subset(qBiCo_data,!qBiCo_data$hematin == 'C') q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$hematin,q_BC$Recovery,'BC'),cbind(q_EC$hematin,q_EC$Recovery,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_rec <- append(stat_list_rec, list(stat_list)) } if (print_stats == TRUE){ #f9 <- f9 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f9 <- f9 + annotate('text',x=0.5, y =y_lim_rec_s[1], label = p_val, vjust='bottom', hjust = 'left')} } } else if (parameter_name == 'inhibition_proteinase') { if (dim(qBiCo_data)[1] == 0) {f10 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$proteinase[grepl('HP',qBiCo_data$Sample)] <- '0.2' qBiCo_data$proteinase[grepl('MP',qBiCo_data$Sample)] <- '0.1' qBiCo_data$proteinase[grepl('LP',qBiCo_data$Sample)] <- '0.05' qBiCo_data$proteinase[grepl('SHS',qBiCo_data$Sample)] <- 'C' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Recovery', groupnames = c('proteinase','conversion_method')) number_levels10 <- c("C","0.05","0.1","0.2") f10 <- ggplot(qBiCo_data_plot,aes(x=factor(proteinase, level = number_levels10), y=Recovery, colour = conversion_method)) f10 <- f10 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f10 <- f10 + coord_cartesian(ylim=y_lim_rec_s) + scale_y_continuous(breaks=y_lim_coord_s) f10 <- f10 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f10 <- f10 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f10 <- f10 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f10 <- f10 + aes(ymax = Recovery + sd, ymin = Recovery - sd) f10 <- f10 + geom_errorbar(aes(ymax = Recovery + sd, ymin = Recovery - sd), position = position_dodge(width = 0.9)) f10 <- f10 + theme(plot.title = element_text(size = 12)) f10 <- f10 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Recovery ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_rec <- append(stat_list_rec, list(stat_list)) qBiCo_data$proteinase[qBiCo_data$proteinase == 'C'] <- '0' q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$proteinase,q_BC$Recovery,'BC'),cbind(q_EC$proteinase,q_EC$Recovery,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_rec <- append(stat_list_rec, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Recovery ~ proteinase, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '0', group.by = 'conversion_method') stat_list_pairwise_rec = append(stat_list_pairwise_rec, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f10 <- f10 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f10 <- f10 + annotate('text',x=0.5, y =y_lim_rec_s[1], label = p_val, vjust='bottom', hjust = 'left')} } } else if (parameter_name == 'freezing') { if (dim(qBiCo_data)[1] == 0) {f11 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$freezing[grepl('5ft',qBiCo_data$Sample)] <- '5' qBiCo_data$freezing[grepl('10ft',qBiCo_data$Sample)] <- '10' qBiCo_data$freezing[grepl('SHS',qBiCo_data$Sample)] <- '0' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Recovery', groupnames = c('freezing','conversion_method')) number_levels11 <- c("0","5","10") f11 <- ggplot(qBiCo_data_plot,aes(x=factor(freezing, level = number_levels11), y=Recovery, colour = conversion_method)) f11 <- f11 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f11 <- f11 + coord_cartesian(ylim=y_lim_rec_s) + scale_y_continuous(breaks=y_lim_coord_s) f11 <- f11 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f11 <- f11 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f11 <- f11 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f11 <- f11 + aes(ymax = Recovery + sd, ymin = Recovery - sd) f11 <- f11 + geom_errorbar(aes(ymax = Recovery + sd, ymin = Recovery - sd), position = position_dodge(width = 0.9)) f11 <- f11 + theme(plot.title = element_text(size = 12)) f11 <- f11 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Recovery ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_rec <- append(stat_list_rec, list(stat_list)) q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$freezing,q_BC$Recovery,'BC'),cbind(q_EC$freezing,q_EC$Recovery,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_rec <- append(stat_list_rec, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Recovery ~ freezing, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '0', group.by = 'conversion_method') stat_list_pairwise_rec = append(stat_list_pairwise_rec, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f11 <- f11 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f11 <- f11 + annotate('text',x=0.5, y =y_lim_rec_s[1], label = p_val, vjust='bottom', hjust = 'left')} } } else {TRUE} } foldername_tiff <- paste0(foldername, '/TIFF') width_tiff <- 3 height_tiff <- 3 ggsave(file = paste0(foldername_tiff,"/Recovery/",'repeatability', ".TIFF"), f1, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Recovery/",'reproducibility', ".TIFF"), f2, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Recovery/",'recovery', ".TIFF"), f3, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Recovery/",'stability_UV', ".TIFF"), f4, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Recovery/",'stability_sonication', ".TIFF"), f5, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Recovery/",'stability_storage', ".TIFF"), f6, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Recovery/",'linearity', ".TIFF"), f7, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Recovery/",'incubation_time', ".TIFF"), f8, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Recovery/",'inhibition_hematin', ".TIFF"), f9, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Recovery/",'inhibition_proteinase', ".TIFF"), f10, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Recovery/",'freezing', ".TIFF"), f11, width = width_tiff, height = height_tiff) title_g <- 'Recovery' y_left <- richtext_grob("**Detected/Expected concentration**", rot=90, gp = gpar(fontsize = 16)) for (figure_lab in c('f1','f2','f3','f8','f7','f4','f5','f9','f10','f6','f11')){ index = which(c('f1','f2','f3','f8','f7','f4','f5','f9','f10','f6','f11')==figure_lab) assign(figure_lab,get(figure_lab) + labs(tag = LETTERS[index])) } g <- grid.arrange(arrangeGrob(f1,f2,f_legend, nrow = 1, widths = c(5,5,1)), arrangeGrob(f3,f8,f7,f4,f5,f9,f10,f6,f11, nrow = 3), nrow = 2, heights = c(2,5), left = y_left) output_g_name <- 'g_rec' assign(output_g_name,g) output_stat_name <- 'stat_rec' assign(output_stat_name,stat_list_rec) #Fragmentation#### y_lim_frag = c(-5,60) y_lim_coord = c(0,10,20,30,40,50,60) y_lim_frag_s = c(-1,15) y_lim_coord_s = c(0,5,10,15) y_lim_frag_lin = c(-1,20) y_lim_coord_lin = c(0,5,10,15,20) x_lim_coord_lin = c(0,25,50,75,100) graph_info_selected <- subset(graph_info,graph_info$index=='fragmentation') #for (instrument_name in subset(graph_info_selected$instrument,!duplicated(graph_info_selected$instrument))) {#create one grob per machine for (parameter_name in c("repeatability" , "reproducibility" ,"recovery", "stability_UV", "stability_sonication", "stability_storage", "linearity", "incubation_time","inhibition_hematin","inhibition_proteinase", "freezing")) {#create one plot per parameter graph_data <- subset(graph_info_selected,graph_info_selected$parameter == parameter_name) x_label = graph_data$x_axis[1] y_label = graph_data$y_axis[1] #select samples/wells #obtain qBiCo indices per well qBiCo_data = df_qBiCo_validation #subset to select only wells that are needed for the plot well_list = graph_data$well run_list = graph_data$run combined_list = paste0(well_list,run_list) qBiCo_data$combined <- paste0(qBiCo_data$Well,qBiCo_data$experiment) qBiCo_data = subset(qBiCo_data,qBiCo_data$combined %in% combined_list) if (parameter_name == 'repeatability') { if (dim(qBiCo_data)[1] == 0){f1 <- ggplot()} #return empty plot if dataframe is empty else{ number_levels1 <- subset(qBiCo_data$DNA_input_ng,!duplicated(qBiCo_data$DNA_input_ng)) #pick amounts for x-axis f1 <- ggplot(qBiCo_data,aes(x=factor(DNA_input_ng, level = number_levels1), y=Fragmentation, color = conversion_method)) f1 <- f1 + geom_boxplot(data=qBiCo_data, outlier.shape = NA,position = position_dodge(preserve = "single"),aes(group = interaction(conversion_method,DNA_input_ng)), color = c("#00BFC4","#00BFC4","#F79B80","#00BFC4","#F79B80","#00BFC4","#F79B80","#00BFC4","#F79B80")) f1 <- f1 + scale_color_manual(values = c("#000000","#000000","#00BFC4","#F79B80")) f1 <- f1 + geom_miss_point(aes(shape = conversion_method,group = conversion_method,colour = interaction(..group..,..missing..)),size = point_size/2,position = position_jitterdodge()) f1 <- f1 + coord_cartesian(ylim=y_lim_frag_s) + scale_y_continuous(breaks=y_lim_coord_s) f1 <- f1 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f1 <- f1 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f1 <- f1 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) f1 <- f1 + theme(plot.title = element_text(size = 12)) f1 <- f1 + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Fragmentation ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_frag <- append(stat_list_frag, list(stat_list)) q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$DNA_input_ng,q_BC$Fragmentation,'BC'),cbind(q_EC$DNA_input_ng,q_EC$Fragmentation,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- anova_test(data = data.bind, formula = y ~ group, covariate = x)#T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p) stat_list_frag = list(stat_list) #pairwise wilcoxon ranksum test (separate per conversion method) f1 <- f1 + stat_compare_means(method = 'wilcox.test', paired = FALSE, label = 'p.signif', hide.ns = TRUE) qBiCo_data <- subset(qBiCo_data,qBiCo_data$DNA_input_ng != '1') qBiCo_data <- subset(qBiCo_data,qBiCo_data$DNA_input_ng != '5') stat_list_pairwise <- compare_means(Fragmentation ~ conversion_method, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'DNA_input_ng') stat_list_pairwise_frag = list(stat_list_pairwise) } if (print_stats_AB == TRUE){ p_val <- paste0('p: ',t1$p) f1 <- f1 + annotate('text',x=0.5, y =-1, label = p_val, vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'reproducibility'){ if (dim(qBiCo_data)[1] == 0) {f2 <- ggplot()} #return empty plot if dataframe is empty else { number_levels2 <- subset(qBiCo_data$DNA_input_ng,!duplicated(qBiCo_data$DNA_input_ng)) #pick amounts for x-axis number_levels2 <- number_levels2[rev(order(number_levels2))] #number_levels2 <- number_levels2[1:minimum_signficicant_conc_index] #only use significantly non different concentrations to compare the 2 runs. qBiCo_data <- subset(qBiCo_data,qBiCo_data$DNA_input_ng %in% number_levels2) qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Fragmentation', groupnames = c('DNA_input_ng','conversion_method','experiment')) f2 <- ggplot(qBiCo_data_plot,aes(x=factor(DNA_input_ng, level = number_levels2), y=Fragmentation, colour = conversion_method, group = experiment)) #f2 <- f2 + facet_grid(cols = vars(factor(DNA_input_ng,level = number_levels2)), scales = 'free', switch = "both") + theme(axis.ticks.x = element_blank(), axis.text.x = element_blank()) f2 <- f2 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f2 <- f2 + coord_cartesian(ylim=y_lim_frag_s) + scale_y_continuous(breaks=y_lim_coord_s) f2 <- f2 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f2 <- f2 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold'), legend.title=element_blank()) f2 <- f2 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) legends <- get_legend(f2) f_legend <- as_ggplot(legends) f2 <- f2 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) f2 <- f2 + geom_errorbar(aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd), position = position_dodge(width = 0.9)) #f2 <- f2 + theme(axis.text.x = element_text(angle = 90, vjust = 0.5, hjust=1)) experiments <- subset(qBiCo_data$experiment,!duplicated(qBiCo_data$experiment)) #qBiCo_data_1 <- subset(qBiCo_data,qBiCo_data$experiment == experiments[1]) #f2 <- f2 + stat_compare_means(data = qBiCo_data_1, label.y = 101, label = 'p.signif', method = 't.test', paired = FALSE, ref.group = '5', hide.ns = hide_ns) #qBiCo_data_2 <- subset(qBiCo_data,qBiCo_data$experiment == experiments[2]) #f2 <- f2 + stat_compare_means(data = qBiCo_data_2, label.y = 100, label = 'p.signif', method = 't.test', paired = FALSE, ref.group = '5', hide.ns = hide_ns) #f2 <- f2 + stat_compare_means(data = qBiCo_data_1, method = 'kruskal.test', label.y = 92, label.x = 2) #f2 <- f2 + stat_compare_means(data = qBiCo_data_2, method = 'kruskal.test', label.y = 90, label.x = 2) f2 <- f2 + theme(plot.title = element_text(size = 12)) #We want to compare the two runs if (perform_stats == TRUE){ f2 <- f2 + stat_compare_means(method = 'wilcox.test', paired = FALSE, label = 'p.signif', aes(group = conversion_method), hide.ns = TRUE) #stat_list <- compare_means(Fragmentation ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_frag <- append(stat_list_frag, list(stat_list)) q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$DNA_input_ng,q_BC$Fragmentation,'BC'),cbind(q_EC$DNA_input_ng,q_EC$Fragmentation,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- anova_test(data = data.bind, formula = y ~ group, covariate = x)#T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p) stat_list_frag <- append(stat_list_frag, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) qBiCo_data <- subset(qBiCo_data,qBiCo_data$DNA_input_ng != '1') qBiCo_data <- subset(qBiCo_data,qBiCo_data$DNA_input_ng != '5') stat_list_pairwise <- compare_means(Fragmentation ~ conversion_method, data= qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'DNA_input_ng') stat_list_pairwise_frag = append(stat_list_pairwise_frag, list(stat_list_pairwise)) } if (print_stats_AB == TRUE){ p_val <- paste0('p: ',t1$p) f2 <- f2 + annotate('text',x=0.5, y =-1, label = p_val, vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'recovery') { if (dim(qBiCo_data)[1] == 0) {f3 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$rec_method[grepl('binding5',qBiCo_data$Sample)] <- '5m' qBiCo_data$rec_method[grepl('binding1',qBiCo_data$Sample)] <- '1m' qBiCo_data$rec_method[grepl('double',qBiCo_data$Sample)] <- '2x' qBiCo_data$rec_method[grepl('SHS',qBiCo_data$Sample)] <- 'C' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Fragmentation', groupnames = c('rec_method','conversion_method')) number_levels3 <- c("C","1m","5m",'2x') f3 <- ggplot(qBiCo_data_plot,aes(x=factor(rec_method, level = number_levels3), y=Fragmentation)) f3 <- f3 + geom_miss_point(aes(shape = conversion_method, colour = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f3 <- f3 + facet_grid(.~ conversion_method, scales = 'free') + theme( strip.background = element_blank(), strip.text.x = element_blank()) f3 <- f3 + coord_cartesian(ylim=y_lim_frag_s) + scale_y_continuous(breaks=y_lim_coord_s) f3 <- f3 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f3 <- f3 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f3 <- f3 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f3 <- f3 + aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd) f3 <- f3 + geom_errorbar(aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd, colour = conversion_method), position = position_dodge(width = 0.9)) f3 <- f3 + theme(plot.title = element_text(size = 12)) f3 <- f3 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ BC_pairwise <- compare_means(Fragmentation ~ rec_method, data = subset(qBiCo_data,qBiCo_data$conversion_method == 'BC'), method = 'kruskal.test') EC_pairwise <- compare_means(Fragmentation ~ rec_method, data = subset(qBiCo_data,qBiCo_data$conversion_method == 'EC'), method = 'kruskal.test') BC_pairwise$conversion_method <- 'BC' EC_pairwise$conversion_method <- 'EC' pairwise_tests <- rbind(BC_pairwise,EC_pairwise) stat_list_frag <- append(stat_list_frag, list(c('Kruskal-Wallis',BC_pairwise$p,'BC'))) stat_list_frag <- append(stat_list_frag, list(c('Kruskal-Wallis',EC_pairwise$p,'EC'))) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Fragmentation ~ rec_method, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = 'C', group.by = 'conversion_method') stat_list_pairwise_frag = append(stat_list_pairwise_frag, list(stat_list_pairwise))} if (print_stats == TRUE){ #f3 <- f3 + stat_compare_means( data = qBiCo_data, method = 'kruskal.test', hide.ns = hide_ns) data_text <- data.frame(label <- paste('p: ',round(pairwise_tests$p,5)), conversion_method <- c('BC','EC')) colnames(data_text) <- c('label','conversion_method') f3 <- f3 + geom_text(data = data_text,mapping = aes(x=0.5, y =y_lim_frag_s[1],label = label) , vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'stability_UV') { if (dim(qBiCo_data)[1] == 0) {f4 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$UV[grepl('30',qBiCo_data$Sample)] <- '30' qBiCo_data$UV[grepl('60',qBiCo_data$Sample)] <- '60' qBiCo_data$UV[grepl('120',qBiCo_data$Sample)] <- '120' qBiCo_data$UV[grepl('SHS',qBiCo_data$Sample)] <- '0' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Fragmentation', groupnames = c('UV','conversion_method')) number_levels4 <- c("0","30","60","120") f4 <- ggplot(qBiCo_data_plot,aes(x=factor(UV, level = number_levels4), y=Fragmentation, colour = conversion_method)) f4 <- f4 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f4 <- f4 + coord_cartesian(ylim=y_lim_frag) + scale_y_continuous(breaks=y_lim_coord) f4 <- f4 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f4 <- f4 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f4 <- f4 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f4 <- f4 + aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd) f4 <- f4 + geom_errorbar(aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd, colour = conversion_method), position = position_dodge(width = 0.9)) f4 <- f4 + theme(plot.title = element_text(size = 12)) f4 <- f4 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Fragmentation ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'UV') #stat_list_frag <- append(stat_list_frag, list(stat_list)) #ANCOVA # res.aov <- qBiCo_data %>% anova_test(Fragmentation ~ UV + conversion_method) # p_val <- paste0('p: ',res.aov$p[2]) # f4 <- f4 + annotate('text',x=0.5, y =y_lim_frag[1], label = p_val, vjust='bottom', hjust = 'left') #Non parametric ANCOVA q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$UV,q_BC$Fragmentation,'BC'),cbind(q_EC$UV,q_EC$Fragmentation,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) p_val <- paste0('p: ',round(t1$p.value,5)) stat_list <- c('p: ',t1$p.value) stat_list_frag <- append(stat_list_frag, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Fragmentation ~ UV, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '0', group.by = 'conversion_method') stat_list_pairwise_frag = append(stat_list_pairwise_frag, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f5 <- f5 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f4 <- f4 + annotate('text',x=0.5, y =y_lim_frag[1], label = p_val, vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'stability_sonication') { if (dim(qBiCo_data)[1] == 0) {f5 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$SON[grepl('SON150',qBiCo_data$Sample)] <- '150' qBiCo_data$SON[grepl('SON500',qBiCo_data$Sample)] <- '500' qBiCo_data$SON[grepl('SON1000',qBiCo_data$Sample)] <- '1000' qBiCo_data$SON[grepl('SON0',qBiCo_data$Sample)] <- 'C' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Fragmentation', groupnames = c('SON','conversion_method')) number_levels5 <- c("C","1000","500","150") f5 <- ggplot(qBiCo_data_plot,aes(x=factor(SON, level = number_levels5), y=Fragmentation, colour = conversion_method)) f5 <- f5 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f5 <- f5 + coord_cartesian(ylim=y_lim_frag) + scale_y_continuous(breaks=y_lim_coord) f5 <- f5 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f5 <- f5 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f5 <- f5 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f5 <- f5 + aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd) f5 <- f5 + geom_errorbar(aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd), position = position_dodge(width = 0.9)) f5 <- f5 + theme(plot.title = element_text(size = 12)) f5 <- f5 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Fragmentation ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_frag <- append(stat_list_frag, list(stat_list)) qBiCo_data$SON[qBiCo_data$SON == 'C'] <- '2000' q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$SON,q_BC$Fragmentation,'BC'),cbind(q_EC$SON,q_EC$Fragmentation,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_frag <- append(stat_list_frag, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Fragmentation ~ SON, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '2000', group.by = 'conversion_method') stat_list_pairwise_frag = append(stat_list_pairwise_frag, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f5 <- f5 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f5 <- f5 + annotate('text',x=0.5, y =y_lim_frag[1], label = p_val, vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'stability_storage') { if (dim(qBiCo_data)[1] == 0) {f6 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$store[grepl('BC_2',qBiCo_data$experiment)] <- 'No storage' qBiCo_data$store[grepl('BC_7',qBiCo_data$experiment)] <- '4 weeks' qBiCo_data$store[grepl('EC_7',qBiCo_data$experiment)] <- 'No storage' qBiCo_data$store[grepl('EC_5',qBiCo_data$experiment)] <- '4 weeks' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Fragmentation', groupnames = c('store','conversion_method')) number_levels6 <- c("No storage","4 weeks") f6 <- ggplot(qBiCo_data_plot,aes(x=factor(store, level = number_levels6), y=Fragmentation, colour = conversion_method)) f6 <- f6 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f6 <- f6 + coord_cartesian(ylim=y_lim_frag_s) + scale_y_continuous(breaks=y_lim_coord_s) f6 <- f6 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f6 <- f6 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f6 <- f6 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f6 <- f6 + aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd) f6 <- f6 + geom_errorbar(aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd), position = position_dodge(width = 0.9)) f6 <- f6 + theme(plot.title = element_text(size = 12)) f6 <- f6 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Fragmentation ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_frag <- append(stat_list_frag, list(stat_list)) qBiCo_data$store[qBiCo_data$store == 'No storage'] <- '0' qBiCo_data$store[qBiCo_data$store == '4 weeks'] <- '4' q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$store,q_BC$Fragmentation,'BC'),cbind(q_EC$store,q_EC$Fragmentation,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_frag <- append(stat_list_frag, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Fragmentation ~ store, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '0', group.by = 'conversion_method') stat_list_pairwise_frag = append(stat_list_pairwise_frag, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f6 <- f6 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f6 <- f6 + annotate('text',x=0.5, y =y_lim_frag_s[1], label = p_val, vjust='bottom', hjust = 'left')} } } else if (parameter_name == 'linearity') { if (dim(qBiCo_data)[1] == 0) {f7 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$Meth_perc[grepl('M0',qBiCo_data$Sample)] <- 0 qBiCo_data$Meth_perc[grepl('M25',qBiCo_data$Sample)] <- 25 qBiCo_data$Meth_perc[grepl('M50',qBiCo_data$Sample)] <- 50 qBiCo_data$Meth_perc[grepl('M75',qBiCo_data$Sample)] <- 75 qBiCo_data$Meth_perc[grepl('M100',qBiCo_data$Sample)] <- 100 qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Fragmentation', groupnames = c('Meth_perc','conversion_method')) number_levels7 <- c(0,25,50,75,100) f7 <- ggplot(qBiCo_data_plot,aes(x=factor(Meth_perc, level = number_levels7), y=Fragmentation, colour = conversion_method)) f7 <- f7 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f7 <- f7 + coord_cartesian(ylim=y_lim_frag_s) + scale_y_continuous(breaks=y_lim_coord_s) f7 <- f7 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f7 <- f7 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f7 <- f7 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f7 <- f7 + aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd) f7 <- f7 + geom_errorbar(aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd), position = position_dodge(width = 0.9)) f7 <- f7 + theme(plot.title = element_text(size = 12)) f7 <- f7 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Fragmentation ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_frag <- append(stat_list_frag, list(stat_list)) q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$Meth_perc,q_BC$Fragmentation,'BC'),cbind(q_EC$Meth_perc,q_EC$Fragmentation,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_frag <- append(stat_list_frag, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Fragmentation ~ Meth_perc, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '100', group.by = 'conversion_method') stat_list_pairwise_frag = append(stat_list_pairwise_frag, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f7 <- f7 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f7 <- f7 + annotate('text',x=0.5, y =-1, label = p_val, vjust='bottom', hjust = 'left')} } } else if (parameter_name == 'incubation_time') { if (dim(qBiCo_data)[1] == 0) {f8 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$incub_time <- 'C' qBiCo_data$incub_time[grepl('12h',qBiCo_data$Sample)] <- '-' qBiCo_data$incub_time[grepl('20h',qBiCo_data$Sample)] <- '+' qBiCo_data$incub_time[grepl('APOBEC',qBiCo_data$Sample)] <- '2+' qBiCo_data$incub_time[grepl('TET',qBiCo_data$Sample)] <- '1+' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Fragmentation', groupnames = c('incub_time','conversion_method')) number_levels8 <- c("C",'-','+','1+', '2+') f8 <- ggplot(qBiCo_data_plot,aes(x=factor(incub_time, level = number_levels8), y=Fragmentation)) f8 <- f8 + geom_miss_point(aes(shape = conversion_method, colour = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f8 <- f8 + facet_grid(.~ conversion_method, scales = 'free') + theme( strip.background = element_blank(), strip.text.x = element_blank()) f8 <- f8 + coord_cartesian(ylim=y_lim_frag_s) + scale_y_continuous(breaks=y_lim_coord_s) f8 <- f8 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f8 <- f8 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f8 <- f8 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f8 <- f8 + aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd) f8 <- f8 + geom_errorbar(aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd, colour = conversion_method), position = position_dodge(width = 0.9)) f8 <- f8 + theme(plot.title = element_text(size = 12)) f8 <- f8 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ BC_pairwise <- compare_means(Fragmentation ~ incub_time, data = subset(qBiCo_data,qBiCo_data$conversion_method == 'BC'), method = 'kruskal.test') EC_pairwise <- compare_means(Fragmentation ~ incub_time, data = subset(qBiCo_data,qBiCo_data$conversion_method == 'EC'), method = 'kruskal.test') stat_list_frag <- append(stat_list_frag, list(c('Kruskal-Wallis',BC_pairwise$p,'BC'))) stat_list_frag <- append(stat_list_frag, list(c('Kruskal-Wallis',EC_pairwise$p,'EC'))) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Fragmentation ~ incub_time, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = 'C', group.by = 'conversion_method') stat_list_pairwise_frag = append(stat_list_pairwise_frag, list(stat_list_pairwise))} if (print_stats == TRUE){ #f8 <- f8 + stat_compare_means(aes(colour = experiment),data = subset(qBiCo_data,qBiCo_data$conversion_method == 'BC'), method = 'kruskal.test', hide.ns = hide_ns) #f8 <- f8 + stat_compare_means(aes(colour = experiment),data = subset(qBiCo_data,qBiCo_data$conversion_method == 'EC'), method = 'kruskal.test', hide.ns = hide_ns) data_text <- data.frame(label <- paste('p: ',round(pairwise_tests$p,5)), conversion_method <- c('BC','EC')) colnames(data_text) <- c('label','conversion_method') f8 <- f8 + geom_text(data = data_text,mapping = aes(x=0.5, y =y_lim_frag_s[1],label = label) , vjust='bottom', hjust = 'left') } } } else if (parameter_name == 'inhibition_hematin') { if (dim(qBiCo_data)[1] == 0) {f9 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$hematin[grepl('HH',qBiCo_data$Sample)] <- '200' qBiCo_data$hematin[grepl('MH',qBiCo_data$Sample)] <- '100' qBiCo_data$hematin[grepl('NH',qBiCo_data$Sample)] <- '0' qBiCo_data$hematin[grepl('SHS',qBiCo_data$Sample)] <- 'C' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Fragmentation', groupnames = c('hematin','conversion_method')) number_levels9 <- c("C","0","100","200") f9 <- ggplot(qBiCo_data_plot,aes(x=factor(hematin, level = number_levels9), y=Fragmentation, colour = conversion_method)) f9 <- f9 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f9 <- f9 + coord_cartesian(ylim=y_lim_frag_s) + scale_y_continuous(breaks=y_lim_coord_s) f9 <- f9 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f9 <- f9 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f9 <- f9 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f9 <- f9 + aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd) f9 <- f9 + geom_errorbar(aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd), position = position_dodge(width = 0.9)) f9 <- f9 + theme(plot.title = element_text(size = 12)) f9 <- f9 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Fragmentation ~ hematin, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = 'C', group.by = 'conversion_method') stat_list_pairwise_frag = append(stat_list_pairwise_frag, list(stat_list_pairwise)) #stat_list <- compare_means(Fragmentation ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_frag <- append(stat_list_frag, list(stat_list)) qBiCo_data <- subset(qBiCo_data,!qBiCo_data$hematin == 'C') q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$hematin,q_BC$Fragmentation,'BC'),cbind(q_EC$hematin,q_EC$Fragmentation,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_frag <- append(stat_list_frag, list(stat_list)) } if (print_stats == TRUE){ #f9 <- f9 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f9 <- f9 + annotate('text',x=0.5, y =y_lim_frag_s[1], label = p_val, vjust='bottom', hjust = 'left')} } } else if (parameter_name == 'inhibition_proteinase') { if (dim(qBiCo_data)[1] == 0) {f10 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$proteinase[grepl('HP',qBiCo_data$Sample)] <- '0.2' qBiCo_data$proteinase[grepl('MP',qBiCo_data$Sample)] <- '0.1' qBiCo_data$proteinase[grepl('LP',qBiCo_data$Sample)] <- '0.05' qBiCo_data$proteinase[grepl('SHS',qBiCo_data$Sample)] <- 'C' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Fragmentation', groupnames = c('proteinase','conversion_method')) number_levels10 <- c("C","0.05","0.1","0.2") f10 <- ggplot(qBiCo_data_plot,aes(x=factor(proteinase, level = number_levels10), y=Fragmentation, colour = conversion_method)) f10 <- f10 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f10 <- f10 + coord_cartesian(ylim=y_lim_frag_s) + scale_y_continuous(breaks=y_lim_coord_s) f10 <- f10 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f10 <- f10 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f10 <- f10 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f10 <- f10 + aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd) f10 <- f10 + geom_errorbar(aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd), position = position_dodge(width = 0.9)) f10 <- f10 + theme(plot.title = element_text(size = 12)) f10 <- f10 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Fragmentation ~ conversion_method ,data = qBiCo_data, method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_frag <- append(stat_list_frag, list(stat_list)) qBiCo_data$proteinase[qBiCo_data$proteinase == 'C'] <- '0' q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$proteinase,q_BC$Fragmentation,'BC'),cbind(q_EC$proteinase,q_EC$Fragmentation,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_frag <- append(stat_list_frag, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Fragmentation ~ proteinase, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '0', group.by = 'conversion_method') stat_list_pairwise_frag = append(stat_list_pairwise_frag, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f10 <- f10 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f10 <- f10 + annotate('text',x=0.5, y =y_lim_frag_s[1], label = p_val, vjust='bottom', hjust = 'left')} } } else if (parameter_name == 'freezing') { if (dim(qBiCo_data)[1] == 0) {f11 <- ggplot()} #return empty plot if dataframe is empty else { qBiCo_data$freezing[grepl('5ft',qBiCo_data$Sample)] <- '5' qBiCo_data$freezing[grepl('10ft',qBiCo_data$Sample)] <- '10' qBiCo_data$freezing[grepl('SHS',qBiCo_data$Sample)] <- '0' qBiCo_data_plot <- data_summary(qBiCo_data, varname = 'Fragmentation', groupnames = c('freezing','conversion_method')) number_levels11 <- c("0","5","10") f11 <- ggplot(qBiCo_data_plot,aes(x=factor(freezing, level = number_levels11), y=Fragmentation, colour = conversion_method)) f11 <- f11 + geom_miss_point(aes(shape = conversion_method),size = point_size, position = position_dodge(width = 0.9)) f11 <- f11 + coord_cartesian(ylim=y_lim_frag_s) + scale_y_continuous(breaks=y_lim_coord_s) f11 <- f11 + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) f11 <- f11 + labs(x = x_label, y = element_blank()) + theme(text = element_text(face = 'bold')) f11 <- f11 + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) #f11 <- f11 + aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd) f11 <- f11 + geom_errorbar(aes(ymax = Fragmentation + sd, ymin = Fragmentation - sd), position = position_dodge(width = 0.9)) f11 <- f11 + theme(plot.title = element_text(size = 12)) f11 <- f11 + scale_color_manual(values = colours_conversion_method) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) if (perform_stats == TRUE){ #stat_list <- compare_means(Fragmentation ~ conversion_method ,data = qBiCo_data, conversion_method = 'wilcox.test', paired = FALSE, group.by = 'SON') #stat_list_frag <- append(stat_list_frag, list(stat_list)) q_BC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'BC') q_EC <- subset(qBiCo_data,qBiCo_data$conversion_method == 'EC') data.bind <- rbind(cbind(q_BC$freezing,q_BC$Fragmentation,'BC'),cbind(q_EC$freezing,q_EC$Fragmentation,'EC')) data.bind <- data.frame(data.bind) colnames(data.bind)=c('x','y','group') data.bind$x <- as.numeric(paste(data.bind$x)) data.bind$y <- as.numeric(paste(data.bind$y)) data.bind$group <- as.numeric(data.bind$group) data.bind <- subset(data.bind,rowSums(is.na(data.bind)) == 0) t1 <- T.aov(data.bind$x, data.bind$y, data.bind$group) stat_list <- c('p: ',t1$p.value) stat_list_frag <- append(stat_list_frag, list(stat_list)) #pairwise wilcoxon ranksum test (separate per conversion method) stat_list_pairwise <- compare_means(Fragmentation ~ freezing, data = qBiCo_data, method = 'wilcox.test', paired = FALSE, ref.group = '0', group.by = 'conversion_method') stat_list_pairwise_frag = append(stat_list_pairwise_frag, list(stat_list_pairwise)) } if (print_stats == TRUE){ #f11 <- f11 + stat_compare_means(data = qBiCo_data,label.y = 100, method = 'wilcox.test', paired = FALSE, hide.ns = hide_ns, label = "p.signif") p_val <- paste0('p: ',round(t1$p.value,5)) f11 <- f11 + annotate('text',x=0.5, y =y_lim_frag_s[1], label = p_val, vjust='bottom', hjust = 'left')} } } else {TRUE} } foldername_tiff <- paste0(foldername, '/TIFF') width_tiff <- 3 height_tiff <- 3 ggsave(file = paste0(foldername_tiff,"/Fragmentation/",'repeatability', ".TIFF"), f1, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Fragmentation/",'reproducibility', ".TIFF"), f2, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Fragmentation/",'recovery', ".TIFF"), f3, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Fragmentation/",'stability_UV', ".TIFF"), f4, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Fragmentation/",'stability_sonication', ".TIFF"), f5, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Fragmentation/",'stability_storage', ".TIFF"), f6, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Fragmentation/",'linearity', ".TIFF"), f7, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Fragmentation/",'incubation_time', ".TIFF"), f8, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Fragmentation/",'inhibition_hematin', ".TIFF"), f9, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Fragmentation/",'inhibition_proteinase', ".TIFF"), f10, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/Fragmentation/",'freezing', ".TIFF"), f11, width = width_tiff, height = height_tiff) title_g <- 'Fragmentation' y_left <- richtext_grob("**Fragmentation index**", rot=90, gp = gpar(fontsize = 16)) for (figure_lab in c('f1','f2','f3','f8','f7','f4','f5','f9','f10','f6','f11')){ index = which(c('f1','f2','f3','f8','f7','f4','f5','f9','f10','f6','f11')==figure_lab) assign(figure_lab,get(figure_lab) + labs(tag = LETTERS[index])) } g <- grid.arrange(arrangeGrob(f1,f2,f_legend, nrow = 1, widths = c(5,5,1)), arrangeGrob(f3,f8,f7,f4,f5,f9,f10,f6,f11, nrow = 3), nrow = 2, heights = c(2,5), left = y_left) output_g_name <- 'g_frag' assign(output_g_name,g) output_stat_name <- 'stat_frag' assign(output_stat_name,stat_list_frag) #BC_EC_22_gDNA#### df_qbico_22 <- subset(df_qbico_22,!df_qbico_22$Sample %in% c('BC_neg','EC_neg')) #remove negative controls as they only have NA values as expected. y_lim_conv = c(90,102) y_lim_coord_conv = c(90,95,100) f_conv <- ggplot(df_qbico_22,aes(x=conversion_method,y=Conversion_efficiency)) + geom_boxplot(outlier.shape = NA, color = colours_conversion_method) + scale_color_manual(values = c("#000000","#000000","#00BFC4","#F79B80")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) + geom_miss_point(aes(shape = conversion_method,group = conversion_method,colour = interaction(..group..,..missing..)),size = point_size/2, position = position_jitterdodge()) + coord_cartesian(ylim=y_lim_conv) + scale_y_continuous(breaks=y_lim_coord_conv) + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) + labs(x = element_blank(), y = 'Conversion efficiency (%)') + theme(text = element_text(face = 'bold')) + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) + theme(plot.title = element_text(size = 12)) if (perform_stats == TRUE){ #pairwise wilcoxon ranksum test (separate per conversion method) f_conv <- f_conv + stat_compare_means(method = 'wilcox.test', paired = FALSE, label = 'p.signif', hide.ns = TRUE) stat_list_pairwise <- compare_means(Conversion_efficiency ~ conversion_method, data = df_qbico_22, method = 'wilcox.test', paired = FALSE) stat_list_pairwise_gDNA10ng = data.frame(stat_list_pairwise) } y_lim_rec = c(-0.5,5) y_lim_coord_rec = c(0,1,2,3,4,5) f_rec <- ggplot(df_qbico_22,aes(x=conversion_method,y=Recovery)) + geom_boxplot(outlier.shape = NA, color = colours_conversion_method) + scale_color_manual(values = c("#000000","#000000","#00BFC4","#F79B80")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) + geom_miss_point(aes(shape = conversion_method,group = conversion_method,colour = interaction(..group..,..missing..)),size = point_size/2, position = position_jitterdodge()) + coord_cartesian(ylim=y_lim_rec) + scale_y_continuous(breaks=y_lim_coord_rec) + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) + labs(x = element_blank(), y = 'Detected/Expected concentration') + theme(text = element_text(face = 'bold')) + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) + theme(plot.title = element_text(size = 12)) if (perform_stats == TRUE){ #pairwise wilcoxon ranksum test (separate per conversion method) f_rec <- f_rec + stat_compare_means(method = 'wilcox.test', paired = FALSE, label = 'p.signif', hide.ns = TRUE) stat_list_pairwise <- compare_means(Recovery ~ conversion_method, data = df_qbico_22, method = 'wilcox.test', paired = FALSE) stat_list_pairwise_gDNA10ng <- rbind(stat_list_pairwise_gDNA10ng,data.frame(stat_list_pairwise)) } y_lim_frag = c(-1,15) y_lim_coord_frag = c(0,5,10,15) f_frag <- ggplot(df_qbico_22,aes(x=conversion_method,y=Fragmentation)) + geom_boxplot(outlier.shape = NA, color = colours_conversion_method) + scale_color_manual(values = c("#000000","#000000","#00BFC4","#F79B80")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), panel.background = element_blank(), axis.line = element_line(colour = "black"), panel.border = element_rect(colour = "black", fill=NA, size=1)) + geom_miss_point(aes(shape = conversion_method,group = conversion_method,colour = interaction(..group..,..missing..)),size = point_size/2, position = position_jitterdodge()) + coord_cartesian(ylim=y_lim_frag) + scale_y_continuous(breaks=y_lim_coord_frag) + theme(axis.text=element_text(size=10, face = "bold"),axis.title=element_text(size=10,face="bold")) + labs(x = element_blank(), y = 'Fragmentation index') + theme(text = element_text(face = 'bold')) + theme(legend.position = "none") + theme(axis.line = element_line(size = 0.5, colour = "black", linetype=1)) + theme(plot.title = element_text(size = 12)) if (perform_stats == TRUE){ #pairwise wilcoxon ranksum test (separate per conversion method) f_frag <- f_frag + stat_compare_means(method = 'wilcox.test', paired = FALSE, label = 'p.signif', hide.ns = TRUE) stat_list_pairwise <- compare_means(Fragmentation ~ conversion_method, data = df_qbico_22, method = 'wilcox.test', paired = FALSE) stat_list_pairwise_gDNA10ng <- rbind(stat_list_pairwise_gDNA10ng,data.frame(stat_list_pairwise)) } #Manova to compare BC set with EC set combining all three qBiCo indices. res.man <- manova(cbind(Conversion_efficiency,Recovery,Fragmentation) ~ conversion_method, data = df_qbico_22) p_val_manova <- summary(res.man)$stats['conversion_method',"Pr(>F)"] summary.aov(res.man) #Shows which indices differ between the two conversion methods foldername_tiff <- paste0(foldername, '/TIFF') width_tiff <- 3 height_tiff <- 3 ggsave(file = paste0(foldername_tiff,"/BC_EC_22_gDNA/",'Conversion_efficiency', ".TIFF"), f_conv, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/BC_EC_22_gDNA/",'Recovery', ".TIFF"), f_rec, width = width_tiff, height = height_tiff) ggsave(file = paste0(foldername_tiff,"/BC_EC_22_gDNA/",'Fragmentation', ".TIFF"), f_frag, width = width_tiff, height = height_tiff) for (figure_lab in c('f_conv','f_rec','f_frag')){ index = which(c('f_conv','f_rec','f_frag')==figure_lab) assign(figure_lab,get(figure_lab) + labs(tag = LETTERS[index])) } g <- grid.arrange(arrangeGrob(f_conv,f_rec,f_frag, nrow = 1), nrow = 1) output_g_name <- 'g_22gDNA' assign(output_g_name,g) #png_Export#### width_png <- 9 height_png <- 12 foldername_png <- paste0(foldername, '/TIFF') #Export Conversion efficiency ggsave(file = paste0(foldername_png,"/Conversion_efficiency.png"), get('g_conv'), width = width_png, height = height_png, limitsize = FALSE) #Export Recovery ggsave(file = paste0(foldername_png,"/Recovery.png"), get('g_rec'), width = width_png, height = height_png, limitsize = FALSE) #Export Fragmentation ggsave(file = paste0(foldername_png,"/Fragmentation.png"), get('g_frag'), width = width_png, height = height_png, limitsize = FALSE) #Export 22gDNA ggsave(file = paste0(foldername_png,"/22gDNA.png"), get('g_22gDNA'), width = width_png, height = 3, limitsize = FALSE) #Export_qbico_output#### df_qBiCo_validation <- rbind(df_qBiCo_validation,df_qbico_22) df_qBiCo_validation_averaged <- rbind(df_qBiCo_validation_averaged,df_qbico_22_averaged) {write_xlsx(df_qBiCo_validation,paste0(foldername,'/20221230_qBiCo_indices_per_well.xlsx')) write_xlsx(df_qBiCo_validation_averaged,paste0(foldername,'/20221230_qBiCo_indices_per_sample.xlsx')) } #Export_stats#### #pairwise parameter_list <- c("repeatability" , "reproducibility" ,"recovery", "stability_UV", "stability_sonication", "stability_storage", "linearity", "incubation_time","inhibition_hematin","inhibition_proteinase", "freezing") index_list <- c('conv','rec','frag') qindex_list <- c('stat_list_pairwise_conv','stat_list_pairwise_rec','stat_list_pairwise_frag') df_export_stat <- data.frame() for (j in seq_along(index_list)){ for (i in seq_along(parameter_list)){ df_pairwise <- get(qindex_list[j])[[i]] df_pairwise$parameter <- parameter_list[i] df_export_stat <- rbind.fill(df_export_stat,df_pairwise) } } df_export_stat <- df_export_stat[,c(2,10,1,3,4,5,6,7,8,9,11)] df_export_stat <- rbind.fill(df_export_stat,stat_list_pairwise_gDNA10ng) filename <- paste0(foldername, '/pairwise_tests.xlsx') wb <- createWorkbook() sheet_nr <- createSheet(wb) addDataFrame(df_export_stat,sheet = sheet_nr, row.names=FALSE) saveWorkbook(wb,file = filename) #BC vs EC parameter_list <- c("repeatability" , "reproducibility", "recovery", "recovery", "stability_UV", "stability_sonication", "stability_storage", "linearity", "incubation_time", "incubation_time","inhibition_hematin","inhibition_proteinase", "freezing") index_list <- list(stat_list_conv,stat_list_rec,stat_list_frag) index_label_list <- c('Conversion_efficiency', 'Recovery', 'Fragmentation') df_export_stat <- data.frame() for (j in seq_along(index_list)){ for (i in seq_along(parameter_list)){ if (length(index_list[[j]][[i]]) == 3){ list_comp <- c(index_list[[j]][[i]],parameter_list[i],index_label_list[j]) df_comp <- data.frame(list_comp[1],list_comp[2],list_comp[4],list_comp[5],list_comp[3]) colnames(df_comp) <- c('Test','p-value','parameter','index','conversion_method') } else { list_comp <- c(index_list[[j]][[i]],parameter_list[i],index_label_list[j]) df_comp <- data.frame(list_comp[1],list_comp[2],list_comp[3],list_comp[4]) colnames(df_comp) <- c('Test','p-value','parameter','index') } df_export_stat <- rbind.fill(df_export_stat,df_comp) } } df_manova <- data.frame('Manova',p_val_manova,'10ng','all',NA) colnames(df_manova) <- colnames(df_export_stat) df_export_stat <- rbind.fill(df_export_stat,df_manova) filename <- paste0(foldername, '/comparative_tests.xlsx') wb <- createWorkbook() sheet_nr <- createSheet(wb) addDataFrame(df_export_stat,sheet = sheet_nr, row.names=FALSE) saveWorkbook(wb,file = filename)