####################################################################################################
#### Supplementary File 3, Fitness.R ########################################################################
####
#### Fit linear regression on log2ratio of barcode abundance between generations 6 and 20
####
#### Wei Zhao and Celia Payen
#### 2012
####################################################################################################

#### Wrap function for linear spline regression
##

## output contains list of estimated slopes and model p-values
early.linear <- function (inputdir, inputfile, outputdir, output, constant=10, start.tp=6, end.tp=20){
  ## constant: the constant added to each gene number; default is 10
  ## start.tp: the starting time point; default is G6
  ## end.tp: the ending time point; default is G20
  
  ##
  #### Read file and prepare for analysis data
  ##
  expdata <- read.csv(paste(inputdir, inputfile, sep=""))
  timepoints <- colnames(expdata)[-1]  # get the time points
  timepoints <- as.numeric(substr(timepoints,2,5)) # change mode from character to numeric
  timepoints.early <- timepoints[timepoints>=start.tp & timepoints<=end.tp] # get early time points (between G6 and G20)
  
  ## remove genes that have 0 copy number at G0, 
  ## and genes with less than 1 average copy on each time point
  print(paste("Number of genes in", inputfile, ":", dim(expdata)[1])) # original number of genes
  avg.nums <- apply(expdata[,-1], 1, mean, na.rm=T)
  expdata  <- expdata[expdata$G0!=0 & avg.nums>=1,]
  print(paste("Number of genes in", inputfile, "after exclusion of unwanted genes:", dim(expdata)[1]))
  
  ##
  #### Linear regression for each gene
  ##
  
  ## plus a constant to all numbers
  anal.data <- expdata[,-1] + constant
  point.total <- apply(anal.data, 2, sum, na.rm=T) # total numbers for each time point
  
  ## linear regression
  slope <- p.value <- vector(length=dim(expdata)[1])
  library(lmtest) # for likelihood ratio test
  for (i in 1:dim(expdata)[1]){
    ## calculate log2 ratio
    one.gene <- anal.data[i,]
    log2ratio <- as.numeric( log2( one.gene/point.total /as.numeric(one.gene["G0"]/point.total["G0"]) ) )
    log2ratio <- log2ratio[timepoints>=start.tp & timepoints<=end.tp]    
    
    fit <- lm( log2ratio ~ timepoints.early )
    slope[i] <- coef(fit)[2]
    p.value[i] <- lrtest(fit)$P[2] ## p-value from likelihood ratio test
  }
  
  ## output to file
  results1 <- data.frame(genes=expdata[,1], log2ratio)
  results <- data.frame (genes=expdata[,1], slope=slope, p.value=p.value)
  write.csv(results, paste(outputdir, output, "-early-slopes.csv", sep=""))  
  write.csv(results1, paste(outpudir,output,"-log2ratio",sep=""))

}

## Write output (create output folder before running script)  
InputDir <- "/home/Barseq/"
OutputDir <- "/home/Barseq/Fitness/"
## CEN-MoBY 
cen.glu.1 <- early.linear(inputdir=InputDir, inputfile="F8r1.csv", outputdir=OutputDir, output="CEN-GLU-1")
cen.glu.2 <- early.linear(inputdir=InputDir, inputfile="F8r2.csv", outputdir=OutputDir, output="CEN-GLU-2")
cen.pho.1 <- early.linear(inputdir=InputDir, inputfile="F9r1.csv", outputdir=OutputDir, output="CEN-PHO-1")
cen.pho.2 <- early.linear(inputdir=InputDir, inputfile="F9r2.csv", outputdir=OutputDir, output="CEN-PHO-2")
cen.sul.2 <- early.linear(inputdir=InputDir, inputfile="F7r3.csv", outputdir=OutputDir, output="CEN-SUL-1")
cen.sul.3 <- early.linear(inputdir=InputDir, inputfile="F8r3.csv", outputdir=OutputDir, output="CEN-SUL-2")

## 2micron-MoBY
micron.glu.1 <- early.linear(inputdir=InputDir, inputfile="2uF11.csv", outputdir=OutputDir, output="2micron-GLU-1")
micron.glu.2 <- early.linear(inputdir=InputDir, inputfile="2uF7.csv", outputdir=OutputDir, output="2micron-GLU-2")
micron.pho.1 <- early.linear(inputdir=InputDir, inputfile="2uF8.csv", outputdir=OutputDir, output="2micron-PHO-1")
micron.pho.2 <- early.linear(inputdir=InputDir, inputfile="2uF9.csv", outputdir=OutputDir, output="2micron-PHO-2")
micron.sul.1 <- early.linear(inputdir=InputDir, inputfile="2uF6.csv", outputdir=OutputDir, output="2micron-SUL-1")
micron.sul.2 <- early.linear(inputdir=InputDir, inputfile="2uF10.csv", outputdir=OutputDir, output="2micron-SUL-2")

## MM1N
MM1N.glu.1 <- early.linear(inputdir=InputDir, inputfile="MM1NF4.csv", outputdir=OutputDir, output="MM1N-GLU-1")
MM1N.glu.2 <- early.linear(inputdir=InputDir, inputfile="MM1NF5.csv", outputdir=OutputDir, output="MM1N-GLU-2")
MM1N.pho.1 <- early.linear(inputdir=InputDir, inputfile="MM1NF8.csv", outputdir=OutputDir, output="MM1N-PHO-1")
MM1N.pho.2 <- early.linear(inputdir=InputDir, inputfile="MM1NF9.csv", outputdir=OutputDir, output="MM1N-PHO-2")
MM1N.sul.1 <- early.linear(inputdir=InputDir, inputfile="MM1NF1.csv", outputdir=OutputDir, output="MM1N-SUL-1")
MM1N.sul.2 <- early.linear(inputdir=InputDir, inputfile="MM1NF2.csv", outputdir=OutputDir, output="MM1N-SUL-2")

## MM2N
MM2N.glu.1 <- early.linear(inputdir=InputDir, inputfile="MM2NF11.csv", outputdir=OutputDir, output="MM2N-GLU-1")
MM2N.glu.2 <- early.linear(inputdir=InputDir, inputfile="MM2NF7.csv", outputdir=OutputDir, output="MM2N-GLU-2")
MM2N.pho.1 <- early.linear(inputdir=InputDir, inputfile="MM2NF8.csv", outputdir=OutputDir, output="MM2N-PHO-1")
MM2N.pho.2 <- early.linear(inputdir=InputDir, inputfile="MM2NF9.csv", outputdir=OutputDir, output="MM2N-PHO-2")
MM2N.sul.1 <- early.linear(inputdir=InputDir, inputfile="MM2NF6.csv", outputdir=OutputDir, output="MM2N-SUL-1")
MM2N.sul.2 <- early.linear(inputdir=InputDir, inputfile="MM2NF10.csv", outputdir=OutputDir, output="MM2N-SUL-2")