#!/usr/bin/perl -w

# Parsing BLAST reports with BioPerl's Bio::SearchIO module;
# Recovers all HSPs;
# See below for filtering options;

use warnings;
use Bio::SearchIO;

# Prompt the user for the file name if it's not an argument;
# NOTE: BLAST file must be in text (not html) format;
if (! $ARGV[0])
{
   $usage = "./blast_parsing_all.pl BLASTREPORT";
   print "Usage is '$usage' or.....\n";
   print "What is the BLAST file to parse? ";

   # Get input and remove the newline character at the end;
   chomp ($inFile = <STDIN>);
}
else
{
   $inFile = $ARGV[0];
}

$report = new Bio::SearchIO( -file => "$inFile",
              					-format => "blast"   ); 

# Prints the header for the columns;
print "Query_Name\tQuery_Acc\tQuery_Len\tHit_Acc\tHit_Desc\tHSP_rank\t\%_ID\te_Value\tHSP_length\n";


# Go through BLAST reports              
while( $result = $report->next_result ) 
{
   # Go through each matching sequence
   while( $hit = $result->next_hit ) 
   { 
      # Go through each each HSP for the sequence
        while ( $hsp = $hit->next_hsp )
        { 
            # The following "if" statements are used to filter the hits for certain criteria;
            # Remove comments from them and corresponding "{}" to call into effect;
            # Also remember to adjust values to those that you are interested in;
            
#            if ( my $hsp->length('total') > 100 ) 
#            {
#              if ( my $hsp->percent_identity >= 75 )
#              {
#                 if ( my $hsp->evalue >= 0 )
#                 {
            		# Print tab-delimited data about this HSP;
            		print $result->query_name, "\t";
            		print $result->query_accession, "\t";
            		print $result->query_length, "\t";
            		print $hit->accession, "\t";
            		print $hit->description, "\t";
            		print $hsp->rank, "\t";
            		print $hsp->percent_identity, "\t";
            		print $hsp->evalue, "\t";
            		print $hsp->hsp_length, "\n";
#                 }
#              }             
#            }
         } 
    } 
}