R/FGI/2021triUni/do.r

   1 #!/usr/bin/env littler
   2
   3 require(readxl)
   4 require(tibble)
   5
   6 # 设置数值显示位数
   7 options(scipen = 200)
   8
   9 argv <- if (!exists("argv")) commandArgs(TRUE) else as.character(argv)
  10 fpath <- argv[1]
  11 fpattern <- "(NFO|HLA|SNP|MED).xlsx?$"
  12
  13 readXLS <- function(xlsname) {
  14         read <- read_excel(xlsname,col_types="text",col_names=FALSE,.name_repair = ~ LETTERS[seq_along(.x)])
  15         tres <- regexpr('(?<Type>\\w{3}).xls',xlsname,perl=TRUE)
  16         st <- attr(tres, "capture.start")[1, ]
  17         tstr <- toupper(substring(xlsname,st,st + attr(tres, "capture.length")[1, ] -1))
  18         rdat <- as.data.frame(t(column_to_rownames(read,var='A')),row.names=make.names(rep_len(tstr,length(read)-1),unique=TRUE))
  19         return(rdat)
  20 }
  21
  22 analyze <- function(dirname) {
  23         #print(c('Fa',dirname))
  24         fnames <- list.files(file.path(fpath,dirname),pattern=fpattern,ignore.case=TRUE,full.names=TRUE)
  25         NFOname <- fnames[grep('NFO.xls',fnames,ignore.case=TRUE)[1]]
  26         HLAname <- fnames[grep('HLA.xls',fnames,ignore.case=TRUE)[1]]
  27         SNPname <- fnames[grep('SNP.xls',fnames,ignore.case=TRUE)[1]]
  28         MEDname <- fnames[grep('MED.xls',fnames,ignore.case=TRUE)[1]]
  29         fnGot <- paste0(dirname,':N=',NFOname,',S=',SNPname,',H=',HLAname,',M=',MEDname)
  30         readXLS(NFOname)
  31         readXLS(HLAname)
  32         readXLS(SNPname)
  33         reads <- sapply(c(NFOname,HLAname,SNPname),readXLS,USE.NAMES=F)
  34         print(reads)
  35         return(c(fnGot,'---'))
  36 }
  37
  38 if (dir.exists(fpath)) {
  39         flist <- list.dirs(fpath,full.names=FALSE,recursive=FALSE)
  40         #print(flist)
  41         ret <- sapply(flist,analyze)
  42         print(ret)
  43 } else {
  44         cat(paste("[!]Usage: ./do.r <path>\n"))
  45 }
  46