GSM2Sample.pl

use lib 'E:/perl_lib';
use FileSystem;

my $work_dir = 'E:\CEL\GSE15460\GSE15460_RAW\GSE15460.SG.CEL';
my $output_file = 'E:\CEL\GSE15460\GSE15460_RAW\GSE15460.info.txt';

my @CellFiles = ();
@CellFiles = FileSystem::GetFileByPattern($work_dir, '\.CEL', @CellFiles);
foreach $cel (@CellFiles)
{
print "$cel\n";
my @CELContent = FileSystem::ReadFile($cel);
my $sample_info_line = $CELContent[13];
my @sample_info = split(/[\s:]/, $sample_info_line);
print "$sample_info[2]";

FileSystem::WriteFile($output_file, "$cel\t$sample_info[2]\n");


}

FileSystem::Close();

STANDARD TRID Tumors on mRNA (Affy U133P2) Reason for exclusion GEO NGCII011/LGE GC-011LGE-T.CEL Fail QC GSM387788.CEL NGCII035/PCC GC-035PCC-T.CEL Fail QC GSM387797.CEL NGCII038/LYC GC-038LYC-T.CEL Fail QC GSM387798.CEL TGCII021/LAH GC-021LAH-T.CEL Fail QC GSM387790.CEL 980327 GC-980327T.CEL called as "adenosquamous cancer", relapse also adenosquamous GSM387937.CEL 2000619 GC-2000619T.CEL squamous CA GSM387844.CEL TGCII026/GJK GC-026-GJK-T.CEL GIST/squamous GSM387793.CEL TGCII039/TSC GC-039-TSC-T.CEL GIST/squamous GSM387799.CEL