# This is code to replicate the analyses and figures from the article entitled 
# "Impact of sunlight exposure at two larval development stages on the residual efficacy of biolarvicides Bacillus thuringiensis israelensis and Bacillus sphaericus against the main malaria vector Anopheles gambiae s.s." by :
# Barnabas Mahugnon ZOGO; Bertin N’Cho Tchiekoi; Alphonsine A. Koffi; 
# Amal Dahounto; Ludovic P. Ahoua Alou; Lamine Baba-Moussa; 
# Nicolas Moiroux; Cédric Pennetier (2018)
# Code developed by Nicolas Moiroux: nicolas.moiroux@ird.fr
# 

library(dplyr)
library(tidyr)
library(coxme)

path <- "C:/Users/.../" # to modify according to the directory where files have been saved

BTBSWAXexp1 <- read.delim(paste0(path,"BTBSWAXexp1.txt")) # Experiment 1 data (with L2 larvae)
BTBSWAXexp2 <- read.delim(paste0(path,"BTBSWAXexp2.txt")) # Experiment 2 data (with L1 larvae)


# data preparation
BTBSWAXexp1$d_cohort <- as.numeric(substr(BTBSWAXexp1$Days_contact,2,2)) # extract a numeric value of the number of day since larvae have been released in the plot
BTBSWAXexp1$Cohort <- as.factor(BTBSWAXexp1$Cohort)
BTBSWAXexp1$Container<- as.factor(BTBSWAXexp1$Container)
BTBSWAXexp1$N_noEm <- 0
BTBSWAXexp1$Dosage <- relevel(relevel(BTBSWAXexp1$Dosage, "1g/m2"), "Ctrl")

BTBSWAXexp2$d_cohort <- as.numeric(substr(BTBSWAXexp2$Days_contact,2,2)) # extract a numeric value of the number of day since larvae have been released in the plot
BTBSWAXexp2$Cohort <- as.factor(BTBSWAXexp2$Cohort)
BTBSWAXexp2$Container<- as.factor(BTBSWAXexp2$Container)
BTBSWAXexp2$N_noEm <- 0
BTBSWAXexp2$Dosage <- relevel(relevel(BTBSWAXexp2$Dosage, "1g/m2"), "Ctrl")

f_countN_noEm <- function(subBTBSWAX){ # function that calculate the number of larvae that remain to emmerge, each day in each plot (i.e. each line) 
  subBTBSWAX$N_noEm[1] <- subBTBSWAX$N_tested[1] - subBTBSWAX$N_adult[1]
  for (i in 2:nrow(subBTBSWAX)){
    subBTBSWAX$N_noEm[i] <- subBTBSWAX$N_noEm[i-1] - subBTBSWAX$N_adult[i] 
  }
  return(subBTBSWAX)
}

BTBSWAX_data_conversion <- function(BTBSWAX){ # function to convert data to be used for survival analysis
  
  for (c in levels(BTBSWAX$Container)){ # for each plot and each cohort
    for (coh in levels(BTBSWAX$Cohort)){
      subBTBSWAX <- subset(subset(BTBSWAX, Container == c), Cohort == coh ) # create a subset
      subBTBSWAX <- arrange(subBTBSWAX,Days_ttmt)                           # order the subset
      
      subBTBSWAX <- f_countN_noEm(subBTBSWAX) # see function definition 
      #print(subBTBSWAX)
      if (subBTBSWAX$N_noEm[nrow(subBTBSWAX)] < 0 ) { # test if the number N_tested is correct
        subBTBSWAX$N_tested <- subBTBSWAX$N_tested - subBTBSWAX$N_noEm[nrow(subBTBSWAX)] # if not correct, calculation of N_tested
        subBTBSWAX <- f_countN_noEm(subBTBSWAX) # re_calculation of N_noEm
      }
      
      
      g_sub <- gather(subBTBSWAX, status, count, c(N_adult,N_noEm)) # create one line per status (adult / no_em) per day
      g_sub_ad <- g_sub[which((g_sub$count > 0 & g_sub$status == "N_adult")),] # select line (dates) when adult have emmerged
      g_sub_lar <- g_sub[nrow(g_sub),]            # select the line (dates) of the last day that contain the number of larvae that finally never emmerge
      dd <- rbind(g_sub_ad, g_sub_lar)  # compiling of the selection
      dd <- slice(dd,rep(1:n(), dd$count))  # duplicate the line according to the number emmerged each day and number remaining (i.e. dead) at the end of experiment
      dd <- select(dd, -count)              # remove column "count"
      dd <- transform(dd,emm=ifelse(status=="N_adult",1,0), status=NULL)  # change name and value of col status : "emm" with possible value 1 or 0
      # dd has as many rows as the number of larvae tested (i.e. one line per larvae)
      
      #print(nrow(dd))
      if (c == levels(BTBSWAX$Container)[1] && coh == levels(BTBSWAX$Cohort)[1]) {  # aggregate each subset in one final dataset with one line per larvae
        BTBSWAX2 <- dd
      } else {
        BTBSWAX2 <- rbind(BTBSWAX2, dd)
      }
    }
  }
  
  return(BTBSWAX2)
} 

Surv_exp1 <- BTBSWAX_data_conversion(BTBSWAXexp1) # convert data
Surv_exp2 <- BTBSWAX_data_conversion(BTBSWAXexp2)
Surv_exp <- rbind(Surv_exp1,Surv_exp2)


# survival (i.e. emmergence inhibition) analysis
# Mixed effect models

model_m1<-coxme(Surv(d_cohort,emm)~Dosage*Shelter+Days_ttmt+(1|Container), data=Surv_exp1) # exp 1
summary(model_m1)

model_m2<-coxme(Surv(d_cohort,emm)~Dosage*Shelter+Days_ttmt+(1|Container), data=Surv_exp2) # exp 2
summary(model_m2)


###### figures
# survival curve for shelter effect analysis (Experiment 2, L1)
Surv_exp2$DosShelt <- as.factor(paste0(Surv_exp2$Dosage, Surv_exp2$Shelter))
sub1 <- subset(subset(Surv_exp2, Cohort == 1), Dosage == c("Ctrl","2g/m2")) 
model1=survfit(Surv(d_cohort,emm)~DosShelt, data=sub1)  

sub2 <- subset(subset(Surv_exp2, Cohort == 2), Dosage == c("Ctrl","2g/m2")) 
model2=survfit(Surv(d_cohort,emm)~DosShelt, data=sub2)

sub3 <- subset(subset(Surv_exp2, Cohort == 3), Dosage == c("Ctrl","2g/m2")) 
model3=survfit(Surv(d_cohort,emm)~DosShelt, data=sub3)

v1 <- seq
Emmerg <- function(x){ # function used to plot Emergence insetad of emergence inhibition
  y <- 1-x
  return(y)
}

plot(model1,lty=c(2,2,1,1),ylab="Emmergence rate",col=c("black","blue","black","blue"),ylim=c(0,1),xlim=c(0,10),main="", fun = Emmerg ) 
axis(side = 1, at = v1)
plot(model2,lty=c(2,2,1,1),ylab="Emmergence rate",col=c("black","blue","black","blue"),ylim=c(0,1),xlim=c(0,10),main="", fun = Emmerg ) 
axis(side = 1, at = v1)
plot(model3,lty=c(2,2,1,1),ylab="Emmergence rate",col=c("black","blue","black","blue"),ylim=c(0,1),xlim=c(0,10),main="", fun = Emmerg ) 
axis(side = 1, at = v1)


# survival curve for shelter effect analysis (Experiment 1, L2)

Surv_exp1$DosShelt <- as.factor(paste0(Surv_exp1$Dosage, Surv_exp1$Shelter))
sub1 <- subset(subset(Surv_exp1, Cohort == 1), Dosage == c("Ctrl","2g/m2")) 
model1=survfit(Surv(d_cohort,emm)~DosShelt, data=sub1)  

sub2 <- subset(subset(Surv_exp1, Cohort == 2), Dosage == c("Ctrl","2g/m2")) 
model2=survfit(Surv(d_cohort,emm)~DosShelt, data=sub2)

sub3 <- subset(subset(Surv_exp1, Cohort == 3), Dosage == c("Ctrl","2g/m2")) 
model3=survfit(Surv(d_cohort,emm)~DosShelt, data=sub3)
summary(Surv_exp1)
v1 <- seq(0,10)
plot(model1,lty=c(2,2,1,1),ylab="Emmergence inhibition",col=c("black","blue","black","blue"),ylim=c(0,1),xlim=c(0,10),main="", fun = Emmerg ) 
axis(side = 1, at = v1)
plot(model2,lty=c(2,2,1,1),ylab="Emmergence inhibition",col=c("black","blue","black","blue"),ylim=c(0,1),xlim=c(0,10),main="", fun = Emmerg ) 
axis(side = 1, at = v1)
plot(model3,lty=c(2,2,1,1),ylab="Emmergence inhibition",col=c("black","blue","black","blue"),ylim=c(0,1),xlim=c(0,10),main="", fun = Emmerg ) 
axis(side = 1, at = v1)