按照文件要求合并同一样本的多个fastq文件至目标目录。更多知识分享请到 https://zouhua.top/。

脚本

  1. #!/usr/bin/perl
  3. use warnings;
  4. use strict;
  5. use Getopt::Long;
  6. use Cwd 'abs_path';
  8. my ($file, $fq_dir, $out, $help, $version);
  9. GetOptions(
  10.     "f|file:s"  =>  \$file,
  11.     "d|fq_dir:s"   =>  \$fq_dir,
  12.     "o|out:s"   =>  \$out,
  13.     "h|help:s"  =>  \$help,
  14.     "v|version" =>  \$version
  15. );
  16. &usage if(!defined $out);
  18. my $cwd = abs_path;
  19. # output
  20. system "mkdir -p $out" unless(-d $out);
  22. my %sampleid;
  23. open(IN, "$file") or die "can't open $file\n";
  24. <IN>;
  25. while(<IN>){
  26.     chomp;
  27.     my @array = split("\t", $_);
  28.     $sampleid{$array[0]} = 1;
  29. }
  30. close(IN);
  32. my (%read1, %read2);
  33. opendir(DIR, $fq_dir) or die "can't open $fq_dir";
  34. foreach (sort grep(/gz$/, readdir(DIR))){
  35.     my $samplename = $1 if $_ =~ m/(\S+)\_(S\d+)\_L/;    # N47_S181_L001_R1_001.fastq.gz
  36.     if(exists($sampleid{$samplename})){
  37.         my $read_mode = $1 if $_ =~ m/\_(R\d)\_(\d+)\.fastq.gz/;  # N47_S181_L001_R1_001.fastq.gz
  38.         if($read_mode eq "R1"){
  39.             my $sample_key = join("_", $samplename, "R1");
  40.             my $fq1 = join("/", $fq_dir, $_);
  41.             push(@{$read1{$sample_key}}, $fq1);
  42.         }elsif($read_mode eq "R2"){
  43.             my $sample_key = join("_", $samplename, "R2");
  44.             my $fq2 = join("/", $fq_dir, $_);
  45.             push(@{$read2{$sample_key}}, $fq2);
  46.         }
  47.     }
  48. }
  49. closedir(DIR);
  52. my $out_dir = join("/", $cwd, $out);
  54. foreach my $key_r1 (keys %read1){
  55.     my $fq_r1 = join(" ", @{$read1{$key_r1}});
  56.     my $length = scalar(@{$read1{$key_r1}});
  57.     my $target = "$out_dir/$key_r1\.fastq.gz";
  58.     unless(-e $target){
  59.         my $shell = "cat $fq_r1 > $target";
  60.         system("$shell");
  61.         print("$key_r1 ($length fastq) successfully combined\n");
  62.     }
  63. }
  65. foreach my $key_r2 (keys %read2){
  66.     my $fq_r2 = join(" ", @{$read2{$key_r2}});
  67.     my $length2 = scalar(@{$read2{$key_r2}});
  68.     my $target2 = "$out_dir/$key_r2\.fastq.gz";
  69.     unless(-e $target2){
  70.         my $shell2 = "cat $fq_r2 > $target2";
  71.         system("$shell2");
  72.         print("$key_r2 ($length2 fastq) successfully combined\n");
  73.     }
  74. }
  76. sub usage{
  77.         print <<USAGE;
  78. usage:
  79.         perl $0 -f <file> -d <fq_dir> -o <out>
  80. options:
  81.         -f|file     :[essential].
  82.         -d|dir      :[essential].
  83.         -o|out      :[essential].
  84. USAGE
  85.     exit;
  86. };
  88. sub version {
  89.     print <<VERSION;
  90.     version:    v1.0
  91.     update:     20201228 - 20201228
  92.     author:     zouhua1\@outlook.com
  93. VERSION

运行

  1. perl combine_fq.pl -f phenotype.tsv -d /data_backup/rawdata/-o RawData_rename > combine.log