swissport蛋白数据库拆分成不同的子库

swissport蛋白数据库拆分成不同的子库

首先从数据库下载文件

wget ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/taxonomic_divisions/uniprot_sprot_*.dat.gz
wget ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/uniprot_sprot.fasta.gz

使用写好的脚本进行操作

perl split_swissprot.pl
grep ">" uniprot_sprot.fasta |sed 's/>//'|perl -lane 'print qq{$F[0]	}.join(" ",@F[1..$#F])' >uniprot_sprot.id.annot.xls

split_swissprot代码如下

#!/usr/bin/perl -w
use strict;

my $files = `ls taxonomic_divisions/uniprot_sprot_*.dat.gz`;
chomp $files;
my %hash;
open OUT,">swissprot_id.xls";
my @tmp = split(/
/, $files);
for my $id(@tmp){
	chomp $id;
	if($id =~ /uniprot_sprot_(.*).dat.gz/){
		my $class = $1;
		open IN,"gzip -dc $id|" || die $!;
		$/="//
";
		while(<IN>){
			chomp;
			my @array = split(/
/);
			my @array2 = split(/s+/, $array[0]);
			if($array2[0] eq "ID"){
				print OUT "$array2[1]	$class
";
			}
			else{
				print "ID error!";
			}
			$hash{$array2[1]} = $class;
		}
		close IN;
	}
}

open ARCHAEA,">./Archaea.fa";
open BACTERIA,">./Bacteria.fa";
open FUNGI,">./Fungi.fa";
open HUMAN,">./Human.fa";
open INVERTEBRATES,">./Invertebrates.fa";
open MAMMALS,">./Mammals.fa";
open PLANTS,">./Plants.fa";
open RODENTS,">./Rodents.fa";
open VERTEBRATES,">./Vrtebrates.fa";
open VIRUSES,">./Viruses.fa";
open ANIMAL,">./Animal.fa";
open OTHER,">./Other.fa";
open UNKOWN,">./Unkown.fa";

open FASTA,"./uniprot_sprot.fasta";
$/=">";
<FASTA>;
while(<FASTA>){
	chomp;
	my @tmp2 = split(/
/, $_);
	my @tmp3 = split(/s+/, $tmp2[0]);
	if($tmp3[0] =~ /sp|(.*)|(.*)\_(.*)/){
		my $cao = $2."\_".$3;
		unless(exists $hash{$cao}){
			print UNKOWN ">$_";
			print OUT "$cao	unkown
";
			next;
		}
		if($hash{$cao} eq "archaea"){
			print ARCHAEA ">$_";
			print OTHER ">$_";
		}
		elsif($hash{$cao} eq "bacteria"){
			print BACTERIA ">$_";
			print OTHER ">$_";
		}
		elsif($hash{$cao} eq "fungi"){
			print FUNGI ">$_";
			print OTHER ">$_";
		}
		elsif($hash{$cao} eq "human"){
			print HUMAN ">$_";
			print ANIMAL ">$_";
		}
		elsif($hash{$cao} eq "invertebrates"){
			print INVERTEBRATES ">$_";
			print ANIMAL ">$_";
		}
		elsif($hash{$cao} eq "mammals"){
			print MAMMALS ">$_";
			print ANIMAL ">$_";
		}
		elsif($hash{$cao} eq "plants"){
			print PLANTS ">$_";
		}
		elsif($hash{$cao} eq "rodents"){
			print RODENTS ">$_";
			print ANIMAL ">$_";
		}
		elsif($hash{$cao} eq "vertebrates"){
			print VERTEBRATES ">$_";
			print ANIMAL ">$_";
		}
		elsif($hash{$cao} eq "viruses"){
			print VIRUSES ">$_";
			print OTHER ">$_";
		}
		else{
			print UNKOWN ">$_";
		}
	}
}
close FASTA;
close OUT;
my $fas = `ls *.fa *.fasta`;
chomp $fas;
open FORMAT,">formatdb.sh";
my @fas_arr = split(/s+/, $fas);
for my $fas_file(@fas_arr){
	chomp $fas_file;
	print FORMAT "/media/sdb/bio/blast/bin/formatdb -p T -i $fas_file
";
}
close FORMAT;