#!/usr/bin/perl -w

# Check to make sure that one file was given on the command line, if not
# print a correct usage statement.
#

use constant CUTOFF => 1E-10;

if (@ARGV != 1) {
    die "\nUsage: ./worm_blast_parse.pl <fasta file>\n\n";
}
print "Top worm blast hits for my 2-hybrid data\n\n";

# Get the file name and open an instream to it.
#
$file = shift;
open IN, $file or die "Couldn't open file: $file\n";

# Read in the contents of the file
# set the array counter at 0

my @output;  #a new array that will store only what I want to keep for output

while ($line = <IN>) {
   
    #first look for "Query" in order to identify the start of an entry and its query number.
    # Find lines containing 'Query=', store them in the array $query_num
    # and add 1 to the counter.
    
    if ($line =~ /^Query=/) { #this line has the sequence name, grab it and print it out
    	my ($junk, $seqname) = split /\s+/, $line; #we split on one or more space, and print out the second element in the list.
	$output[0] = $seqname; #we need to do this, because we need to use the sequence after the if statement
	   
   } elsif ($line =~ /High-scoring Segment Pairs/){ #this is the start of the table of High score pairs
       $line = <IN>; #read a blank line
       $line = <IN>; #this is the first high scoring pair line
       while ($line !~ /^\s/){ #while I'm not at the end of the table, i.e. the first blank line
	   my @pieces = split /\s+/, $line; #I now have an array of all of the words of my line.
	   
	   # we want the ID, the annotation, and the e value
	   $output[1] = shift @pieces; #get the first element of pieces.
	   $output[3] = $pieces[@pieces - 2]; #we know the e-value is the second to last element of pieces
	   pop @pieces;
	   pop @pieces;
	   pop @pieces; #we'll pop off the last three elements (score, evalue, n).
	   shift @pieces; #remove CExxxx
	   shift @pieces; #remove WBGenexxxxxxx
	   #the only thing left in @pieces is the annotation
	   $output[2] = join " ", @pieces;
	   if ($output[3] < CUTOFF){
	       print join "\t", @output;
	       print "\n";
	   }else{
	       last; #we're not less than the cutoff, so there's no reason to go any further. Everything else
	       # will also be greater.
	   }
	   $line = <IN>;
       }
   }
}

close IN;