提取注释信息

根据丰度从nr 注释结果提取每个基因的注释信息

nr.output
gene_1000002|288	288	1	216	75.0	95	1	72	310	2.5e-25	62	72	86.1	WP_019151153.1	WP_019151153.1 hypothetical protein [Alistipes senegalensis]	62	10	0	0	124.0	86.1
gene_1000002|288	288	1	216	75.0	95	1	72	307	5.5e-25	62	72	86.1	WP_010266016.1	WP_010266016.1 hypothetical protein [Alistipes timonensis]	59	13	0	0	122.9	81.9
gene_1000002|288	288	1	216	75.0	95	1	72	306	7.2e-25	62	72	86.1	WP_064974447.1	WP_064974447.1 hypothetical protein [Alistipes sp. Marseille-P2431]	58	14	0	0	122.5	80.6

Donor3.nr.output.forTA
gene_1000007|1197	2.4e-199	WP_008627072.1	1197
gene_1000007|1197	1.2e-198	WP_008619774.1	1197
gene_1000007|1197	3.4e-193	CCZ03104.1	1197
gene_1000007|1197	7.3e-164	CDB11117.1	1197
gene_1000007|1197	9.5e-164	WP_071148079.1	1197
gene_1000007|1197	2.9e-160	WP_005848635.1	1197
gene_1000007|1197	3.7e-160	WP_005846104.1	1197
gene_1000007|1197	3.7e-160	WP_008671032.1	1197

#!/usr/bin/perl -w
use strict;
use Getopt::Long;
my $usage = <<_USAGE_;
usage : 
version 1.0
getDataFromList.pl -i inputList -l log -td TMP_DONE -d target -p perl dir -q query column -t target column -o output dir -m mark
_USAGE_

my ($inputList, $totalRunLog, $TMP_DONE, $target, $queryColumn, $targetColumn, $perl_dir, $outputDir, $mark);
GetOptions(
	"i=s" => \$inputList,
	"d=s" => \$target,
	"l=s" => \$totalRunLog,
	"td=s" => \$TMP_DONE,
	"q=s" => \$queryColumn,
	"t=s" => \$targetColumn,
	"o=s" => \$outputDir,
	"p=s" => \$perl_dir,
	"m=s" => \$mark
);
die $usage if (!$inputList || !$totalRunLog || !$TMP_DONE || !$target || !$queryColumn || !$targetColumn || !$outputDir|| !$perl_dir);

$queryColumn--;
$targetColumn--;
#my @targets = split/\s+/, $targetLine;
#foreach my $target (@targets){
#	$target =~ /.+\/(.+)\./;
#	my $mark = $1;
	my %target;
	open  (my $t, "$target") || die "$!:$target\n";
	while (<$t>){
		chomp;next if ($_ eq '');s/\r//g;
		my @entries = split/\t/;
		if ($target{$entries[$targetColumn]}){
			push (@{$target{$entries[$targetColumn]}}, $_);
		}else{
			$target{$entries[$targetColumn]}->[0] = $_;
		}
	}
	close $t;

	$inputList =~ /.+\/(.+?)\./;
	my $sample = $1;
	open (my $in, "$inputList" ) || die "$!:$inputList\n";
	open (my $out, ">$outputDir/$sample.$mark.output") || die "$!:$outputDir/$sample.$mark.output\n";
	while (<$in>){
		chomp;next if ($_ eq '');s/\r//g;
		my @entries = split/\t/;
		my $key = $entries[$queryColumn];
		if ($target{$key}){
			foreach my $annot ( @{$target{$key}}){
				print $out "$annot\n";
			}
		}
	}
	close $in;
	close $out;
#}