# ==== Open the files and get sequence

library(ShortRead)


dataDir <- "/home/retureau/Bureau/sampleData/output"
fls <- list.files(dataDir, "fastq$", full=TRUE)  # select two fastq files and store their paths
names(fls) <- sub(".fastq", "", basename(fls))  # name the columns. basename(fls) give the "name.fastq" and sub remove ".fastq".

fq <- lapply(fls, readFastq)   # apply readFastq func for both fastq file (store them in fq)

AvgQty <- function(rFq)
{
  # Calculate the average quality of each read
  rowMeans(as(quality(rFq), "matrix"))
}


Rtrim <- function(fq)
{
  qtyMat <- sapply(fq, AvgQty)

  outputDir <- "/home/retureau/Bureau/sampleData"  # output directory

  # full path + new name for output files
  pairedFile <- c(file.path(outputDir, paste0(names(fq)[1], "_paired.fastq")), file.path(outputDir,  paste0(names(fq)[2], "_paired.fastq")))
  unpairedFile <- c(file.path(outputDir, paste0(names(fq)[1], "_funpaired.fastq")), file.path(outputDir,  paste0(names(fq)[2], "_runpaired.fastq")))

  #  dim(qtyMat)[1]
  good_paired_quality = which(qtyMat[,1]>20 & qtyMat[,2]>20)
  if (length(good_paired_quality) > 0){
    writeFastq(fq[[1]][good_paired_quality], pairedFile[1], mode='w')
    writeFastq(fq[[2]][good_paired_quality], pairedFile[2])
  }
  good_unpaired_qual_1=which(qtyMat[, 1]>20 & qtyMat[,2]<=20)
  if (length(good_unpaired_qual_1) > 0)
    writeFastq(fq[[1]][good_unpaired_qual_1], unpairedFile[1])

  good_unpaired_qual_2=which(qtyMat[,1]<=20 & qtyMat[,2]>20)
  if (length(good_unpaired_qual_2) > 0)
    writeFastq(fq[[2]][good_unpaired_qual_2], unpairedFile[2])
}