#!/usr/bin/perl   
# the infile "query.txt" has multiple blast hits per each query
# the outfile needs to have the first blast hit with the Ref number or the best hit
# the second outfile "reflist.txt" is to get the ref numbers to do batch entrez
$infile = "query.txt"; #output of blast
open (IN, $infile) or die "can't open file: $!";

$outfile = "besthit.txt";
open (OUT, ">$outfile") or die "can't open file: $!";

$ofile = "reflist.txt";
open (NP, ">$ofile") or die "can't open file: $!";

while (<IN>) {
	if ($_ =~ /^Query/) {
		chomp;
		$query = $_;
		$query =~ s/^Query=\s//;
		
	}
	
	elsif ($_ =~ s/^\s+\((\d+)\sletters\)/$1/) {
		chomp;
		$length = $_;
		
	}
	
	elsif ($_ =~ /\|ref\|/) {
		chomp;
		$np = $_;
		until (/^\s+/) {
		$_ = <IN>;
		$line = join (" ", $line, $_);
		}
		# print OUT "$np\n";
		$id = substr $np, 0, 70;
		$score = substr $np, 70, 6;
		$score =~ s/^\s//;
		$evalue = substr $np, 76, 6;
		@id = split(/\|/, $id);
		$gino = $id[1];
		$npno = $id[3];
		$stid = $id[4];
		$stid =~ s/^\s//;
		print NP "$npno\n";
		print OUT "\nQURY: $query\n";
		print OUT "SIZE: $length\n";
		print OUT "GINO: $gino\n";
		print OUT "NPNO: $npno\n";
		print OUT "STID: $stid\n";
		print OUT "SCOR: $score\n";
		print OUT "EVAL: $evalue\n";
	} #elsif end

	elsif ($_ =~ /^gi/) {
		chomp;
		$np = $_;
		until (/^\s+/) {
		$_ = <IN>;
		$line = join (" ", $line, $_);
		}
		# print OUT "$np\n";
		$id = substr $np, 0, 70;
		$score = substr $np, 70, 6;
		$score =~ s/^\s//;
		$evalue = substr $np, 76, 6;
		@id = split(/\|/, $id);
		$gino = $id[1];
		$stid = $id[4];
		$stid =~ s/^\s//;
		print OUT "\nQURY: $query\n";
		print OUT "SIZE: $length\n";
		print OUT "GINO: $gino\n";
		print OUT "STID: $stid\n";
		print OUT "SCOR: $score\n";
		print OUT "EVAL: $evalue\n";
	} #elsif end
			
} #while end
close IN;
close OUT;
close NP;

</readmore>