R Notebook Figure 5c

Please use the same folder as work directory as the data files. 1. Remove anything left in the datafiles

rm(list=ls())

If necesearry please install the following packages:2
if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")
BiocManager::install("clusterProfiler")
BiocManager::install("ReactomePA") 
BiocManager::install("rrvgo")
BiocManager::install("enrichplot")
BiocManager::install("msigdbr")
BiocManager::install("org.Hs.eg.db")
BiocManager::install("AnnotationDbi")
install.packages("ggplot2")
install.packages("glue")
BiocManager::install("fgsea")
install.packages("mulea")

2. Reading in necesearry packages

library(clusterProfiler)
library(ReactomePA)
library(rrvgo)
library(enrichplot)
library(ggplot2)
library(msigdbr)
organism ="org.Hs.eg.db"
library(organism, character.only = TRUE)
library(glue)
library(mulea)
library(tidyverse)
library(AnnotationDbi)
library(scales)

3. Data input and prepearation for enrtichment analysis THe working directory to be ecpected to be the folder of the Rmd file

PPI network propagation output

outcome_uc <- read.csv("uc_only_ppi_10k.txt", sep="\t", row.names = 1)
head(outcome_uc)

Reactome gene annotation file

reactome_ontology <- read_gmt("ReactomePathways.gmt")

reactome_ontology <- reactome_ontology %>% 
rename(ontology_id = "ontology_id",
    ontology_name = "ontology_name",
    list_of_values = "list_of_values")
head(reactome_ontology)      

Filtering onytology for minimum 5 and maximmum 400 elments - excluding really large and really small patheways. We can change theese settings later.

reactome_ontology_filtered <- filter_ontology(gmt = reactome_ontology,
                                        min_nr_of_elements = 5,
                                        max_nr_of_elements = 500)

OmniPath newtwork file

op <- read.csv("OmniPath_27_10_2021_directed.ncol", sep=" ", header = FALSE, row.names = NULL)
head(op)

head(outcome_uc)

filtered_outcome_uc <- outcome_uc[outcome_uc$Z_Count>0,]
head(filtered_outcome_uc)

bg <- unique(c(op$V1,op$V2))

# Translating genes in the background and outcome data to gene symbols

# Translate background genes
bg_symbols <- mapIds(org.Hs.eg.db, keys = bg, column = "SYMBOL", keytype = "UNIPROT", multiVals = "first")

4. Enrichmnet and creating Figure 5c

commonly_afffected_genes_uc <- filtered_outcome_uc[filtered_outcome_uc$Z_Count>100,] 

dim(commonly_afffected_genes_uc)

[1] 95  4

commonly_afffected_uc_genes_symbols <- mapIds(org.Hs.eg.db, keys = row.names(commonly_afffected_genes_uc), 
column = "SYMBOL", keytype = "UNIPROT", multiVals = "first")

ora_model <- ora(gmt = reactome_ontology_filtered, 
                 # Test set variable
                 element_names = commonly_afffected_uc_genes_symbols, 
                 # Background set variable
                 background_element_names = bg_symbols, 
                 # p-value adjustment method
                 p_value_adjustment_method = "eFDR", 
                 # Number of permutations
                 number_of_permutations = 10000,
                 # Number of processor threads to use
                 nthreads = 2, 
                 # Setting a random seed for reproducibility
                 random_seed = 42) 

reactome_results_uc_ppi <- run_test(ora_model)

reactome_results_uc_ppi$GeneRatio <- reactome_results_uc_ppi$nr_common_with_tested_elements / length(commonly_afffected_uc_genes_symbols)
sig_results_uc_ppi <- reactome_results_uc_ppi %>%
    # Rows where the eFDR < 0.05
    filter(eFDR < 0.1) %>%
    # Arrange the rows by the gene ratio for plotting
    arrange(desc(GeneRatio))
sig_results_uc_ppi$ontology_id <- factor(sig_results_uc_ppi$ontology_id,
levels = sig_results_uc_ppi[order(sig_results_uc_ppi$GeneRatio, decreasing = FALSE), "ontology_id"])
sig_results_uc_ppi

write.csv(reactome_results_uc_ppi, "Reactome_UC_PPI_100_cutoff_ora_results.csv", row.names = FALSE)

Label formatting

sig_results_uc_ppi$formed_id <- gsub("_", " ", sig_results_uc_ppi$ontology_id)
sig_results_uc_ppi$formed_id <- factor(sig_results_uc_ppi$formed_id,
                                    levels = sig_results_uc_ppi[order(sig_results_uc_ppi$GeneRatio,
                                    decreasing = FALSE), "formed_id"])

Adding a small amount for the fDR calcuation based on the MULEA calcautions.

sig_results_uc_ppi$eFDR = sig_results_uc_ppi$eFDR +10^-12

dotplot_reactome_uc_ppi <- ggplot(sig_results_uc_ppi[1:20,], aes(x=GeneRatio, y=formed_id)) +
    geom_point(aes(size=nr_common_with_tested_elements, colour=eFDR)) +
    scale_y_discrete(labels = label_wrap(50)) + 
    scale_color_continuous(low="red", high="blue", 
                           limits=c(10^-5, 0.1),
                           trans = "log10", name="Emprical FDR")
                

dotplot_reactome_uc_ppi <- dotplot_reactome_uc_ppi +
    scale_size(limits = c(2,20), range = c(2,15), name = "Number of genes") + 
    xlim(0, 0.23) +
    xlab("Gene Ratio") +
    ylab("Reactome pathways") +
    theme_light() +
    theme(rect=element_rect(fill="white")) +
    theme(panel.background = element_rect(fill = "white")) +
    theme(plot.background = element_rect(fill = "white")) +
    theme(plot.background = element_rect(colour = "white"))+
    theme(axis.title.x = element_text(colour = "black", size = 10)) +
    theme(axis.title.y = element_text(colour = "black", size = 10)) +
    theme(axis.text = element_text(color= "black", size = 10)) +
    theme(axis.line = element_line(color = "black")) +
    theme(legend.background = element_rect(fill ="white")) +
    theme(legend.text = element_text(color="black", size = 10)) +
    theme(legend.title = element_text(color= "black", size = 10)) 

dotplot_reactome_uc_ppi

png(file=glue("Figure5c.png"),width=8, height=8, units="in", res=600)
dotplot_reactome_uc_ppi
dev.off()

null device 
          1