comparison add_expression_HPA.R @ 14:133309fd6875 draft

"planemo upload commit c4edbc8a0ffd86395f088cb5b6592f77db658ac7"

13:0b279190f90d

    cat(paste(sep = "", collapse = " ", c("Column",ncol,"not found in file") ))
    stopQuietly()
  }
}

main = function() {
  
  args = get_args()

  #save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/add_expression_data_HPA/args.rda")

    ids = ids[which(!is.na(ids))]
  }
  check_ensembl_geneids(ids)

  # Read protein atlas
  protein_atlas = args$atlas
  protein_atlas = read_file(protein_atlas, T)

  # Add expression
  output = args$output
  options = strsplit(args$select, ",")[[1]]
  res = add_expression(ids, protein_atlas, options)
  
  # Write output
  if (is.null(res)) {
    write.table("None of the ENSG ids entered can be found in HPA data file",file=output,sep="\t",quote=FALSE,col.names=TRUE,row.names=FALSE)

14:133309fd6875

    cat(paste(sep = "", collapse = " ", c("Column",ncol,"not found in file") ))
    stopQuietly()
  }
}

convert_to_previous_header <- function(options){
    header = c('Gene','description','Evidence','Antibody','RNA tissue specificity','Reliability (IH)','Reliability (IF)','Subcellular location','RNA tissue specific NX','TPM max in non-specific')
    names(header) = c('Gene','description','Evidence','Antibody','RNA tissue category','Reliability (IH)','Reliability (IF)','Subcellular location','RNA TS TPM','TPM max in non-specific')
    options = names(header[which(header %in% options)])
    return(options)
}

main = function() {
  
  args = get_args()

  #save(args,file="/home/dchristiany/proteore_project/ProteoRE/tools/add_expression_data_HPA/args.rda")

    ids = ids[which(!is.na(ids))]
  }
  check_ensembl_geneids(ids)

  # Read protein atlas
  protein_atlas = read_file(args$atlas, T)

  # Add expression
  output = args$output
  options = strsplit(args$select, ",")[[1]]
  if (tail(unlist(strsplit(args$atlas,"/")),1) == "HPA_full_atlas_23-10-2018.tsv"){ 
      options = convert_to_previous_header(options)
  } else {
      options = options[which(options != 'TPM max in non-specific')]
      }
  res = add_expression(ids, protein_atlas, options)
  
  # Write output
  if (is.null(res)) {
    write.table("None of the ENSG ids entered can be found in HPA data file",file=output,sep="\t",quote=FALSE,col.names=TRUE,row.names=FALSE)