Create SCTKE from Rsubread and multiqc

createSCTKE_from_Rsubread_and_multiqc.R

library(singleCellTK)

createSCTKE_from_Rsubread_and_multiqc <- function(matdir, r1multi, r2multi){
  r1annot <- read.table(r1multi, sep="\t", row.names = 1, header=TRUE)
  colnames(r1annot) <- paste0(colnames(r1annot), "_R1")
  rownames(r1annot) <- gsub("_R1$", "", rownames(r1annot))
  r2annot <- read.table(r2multi, sep="\t", row.names = 1, header=TRUE)
  colnames(r2annot) <- paste0(colnames(r2annot), "_R2")
  rownames(r2annot) <- gsub("_R2$", "", rownames(r2annot))
  if(any(rownames(r1annot)!=rownames(r2annot))){
    stop("malformed multiqcdata")
  }
  matlist <- list()
  for (i in c("featureCounts", "fpkm", "fpkmlog", "tpm", "tpmlog", "stats")){
    message(i)
    if(i == "featureCounts"){
      name <- "counts"
    } else {
      name <- i
    }
    for (j in rownames(r1annot)){
      message(j)
      infile <- list.files(path = matdir, pattern=paste0(j,"\\.",i,"$"))
      currfile <- read.table(file.path(matdir, infile), header=F, row.names = 1, sep="\t")
      colnames(currfile) <- j
      if(is.null(matlist[[name]])){
        matlist[[name]] <- currfile
      } else {
        matlist[[name]] <- cbind(matlist[[name]], currfile)
      }
    }  
  }
  
  fullannot <- cbind(r1annot, r2annot, t(matlist[["stats"]]))
  outdata <- createSCE(assayFile = matlist[["counts"]], annotFile = fullannot, inputDataFrames = TRUE, createLogCounts = TRUE)
  for (i in names(matlist)[!(names(matlist) %in% c("counts", "stats"))]){
    assay(outdata, i) <- as.matrix(matlist[[i]])
  }
  return(outdata)
}

latent_capetown_2018_SCE <- createSCTKE_from_Rsubread_and_multiqc(
  matdir = "~/Downloads/t/",
  r1multi = "~/Downloads/latent_capetown_2018_r1_multiqc_report_data/multiqc_fastqc.txt",
  r2multi = "~/Downloads/latent_capetown_2018_r2_multiqc_report_data/multiqc_fastqc.txt")