genome_prepare/script/extract_fa_by_list.pl

#!perl -w

die "Usage:\n\tperl $0 id_list input.fa output.fa\n" if(@ARGV !=3);
my ($id_list,$fa,$out_fa)=@ARGV;
my %ids;
open IN,$id_list||die $!;
while(<IN>){
    chomp;
	next if /^$/;
    $ids{$_}=1;
}
close IN;

open OUT,">$out_fa" || die $!;

if($fa=~/.gz$/){
	open IN,"pigz -p 4 -dc $fa"||die $!;
}else{
	open IN,$fa||die $!;
}
# 改输入符；提取序列会有问题
#$/=">";
#while(<IN>){
#	chomp;
#	next if /^$/;
#	my ($id)=$_=~/^([^\s\t]+)/;
#	print OUT ">$_" if($ids{$id});
#}

my $seqname="";
while (<IN>) {
	chomp;
    next if (/^$/);
	($seqname)=$_=~/>(\S+)/ if($_=~/^>/);
	print OUT $_,"\n" if(exists $ids{$seqname});
}

close IN;
close OUT;