Given CSV output from ParsEval, categorize each comparison as a perfect match, a CDS match, an exon structure match, a UTR structure match, or a non-match.

ctgrz.pl

#!/usr/bin/env perl
use strict;

while(my $line = <STDIN>)
{
  chomp($line);
  my @values = split(/,/, $line);

  my $page = sprintf("%s/%d-%d.html", $values[0], $values[1], $values[2]);
  my $cds_match_coef = $values[33];
  my $utr_match_coef = $values[39];
  my $overall_match_coef = $values[32];
  my $correct_exons = $values[16];
  my $refr_exons = $values[14];
  my $wrong_exons = $values[18];
  my $missing_exons = $values[17];

  my $refr_trans = $values[3];
  my $pred_trans = $values[4];

  my $class;
  if($cds_match_coef == 1 and $utr_match_coef == 1)
  {
    if($overall_match_coef == 1)
    {
      $class = "perfect match";
    }
    else
    {
      $class = "perfect match, mislabeled UTRs";
    }
  }

  elsif($cds_match_coef == 1)
  {
    $class = "CDS match";
  }

  elsif($correct_exons == $refr_exons and $wrong_exons + $missing_exons == 0)
  {
    $class = "exon match";
  }

  elsif($utr_match_coef == 1)
  {
    $class = "UTR match";
  }

  else
  {
    $class = "non-match";
  }

  printf("%s\t%s\t%s\t%s\n", $class, $page, $refr_trans, $pred_trans);
}