#! /usr/bin/perl -w

#This script was written by Xiaodong Bai
#command line usage: #perl ParseBlast.pl Blast_output_file Blast_parsed_output_file
#It reads the output file of Blast search
#gives one file named as adding ".all" to the end of user selected Blast_parsed_output_file
#that store all queries and blast information
#AND, gives another file named as adding ".hits" to the end of user selected Blast_parsed_output_file
#that stores only queries with hits and first three hits and E value, if existed

my $Lines = 0;
my @HitLines;
my @Data;
my @Results;
my @Query;

open INPUT, "$ARGV[0]"
    or die "Cannot open input file: $!";

while (<INPUT>) {
    chomp;
    push @Data, $_;
    $Lines += 1;
    if ($_ =~ s/^Query=\s+?(.+?)$/$1/) {
	push @Query, $1;
    } 
    if ($_ =~ /^Searching..../) {
	push @HitLines, $Lines;
    }
}

close INPUT
    or die "Cannot close input file: $!";

foreach (@HitLines) {
    if ($Data[$_ + 1] =~ /No\s*hits\s*found/) {
	push @Results, $Data[$_ + 1];
    } else {
	my $temp = "";
	for (my $i = 0; $i < 3; $i ++) {
	    if ($Data[$_ + 4 + $i] =~ /^(gi|\w)/) {
	        $temp .= "$Data[$_ + 4 + $i]\n";
	    }
	}
	push @Results, $temp;
    }
}

my $Query = @Query;

my $all = $ARGV[1] . ".all";
open OUTPUT, ">$all"
    or die "Cannot open the file to store all queries with or without hits: $!";

for (my $j = 0; $j < $Query; $j ++) {
    print OUTPUT ">$Query[$j]\n";
    print OUTPUT "$Results[$j]\n\n";
}

close OUTPUT
    or die "Cannot close the file to store all queries with or without hits: $!";

my $hits = $ARGV[1] . ".hits";
open OUTPUT2, ">$hits"
    or die "Cannot open the file to store all queries with hits: $!";

for (my $k = 0; $k < $Query; $k ++) {
    if ($Results[$k] =~ /^(gi|\w)/) {
	print OUTPUT2 ">$Query[$k]\n";
	print OUTPUT2 "$Results[$k]\n\n";
    }
}

close OUTPUT2
    or die "Cannot close the file to store all queries with hits: $!";