annotation_summary

#!/usr/bin/perl

use strict;
use warnings;
use Getopt::Long;

use FindBin;
use lib "$FindBin::RealBin/../lib/";

use GAL::Annotation;
use GAL::List;

#-----------------------------------------------------------------------------
#----------------------------------- MAIN ------------------------------------
#-----------------------------------------------------------------------------

my $usage = "

Synopsis:

annotation_summary feature.gff3 genome.fasta

Description:

This script will summarize the annotations in GFF3 file.

";

my ($help);

my $opt_success = GetOptions('help|h' => \$help,
			      );

if (! $opt_success) {
    print STDERR join ' : ', ('FATAL',
                              'command_line_parse_error',
                              'Use annotation_summary --help to see correct usage');
}

if ($help || !@ARGV) {
 print $usage;
 exit(0);
}

my ($feature_file, $fasta_file) = @ARGV;

my $annotation = GAL::Annotation->new($feature_file,
				      $fasta_file);

my $features = $annotation->features;

#for my $column (qw(seqids sources types strands phases)) {
#  print $features->$column->count_table($column, qw(count));
#}
#
my @types = $features->types->uniq;
#
#for my $method (qw(length gc_content)) {
#  for my $type (@types) {
#    my $type_rs = $features->search({type => $type});
#    my $values = $type_rs->method_list($method);
#    print join "\t", $type, $values->mean;
#    print "\n";
#    print '';
#  }
#  print '';
#}

for my $type (@types) {
  my $type_rs = $features->search({type => $type});
  my %counts;
  while (my $feature = $type_rs->next) {
    my $children = $feature->children;
    while (my $child = $children->next) {
      $counts{$child->type}++;
    }
  }
  next unless %counts;
  print "$type\n";
  for my $child_type (keys %counts) {
    print "$child_type\t" . $counts{$child_type};
    print "\n";
    print '';
  }
  print '';
}
print '';


# Do the search and get an interator for all matching features
my $mrnas = $features->search({type => 'mRNA'});

# Iterate over the features
while (my $mrna = $mrnas->next) {
  # Get the feature ID
  my $mrna_id = $mrna->feature_id;
  # Get all the exons for this mRNA
  my $exons = $mrna->exons;
  my $e_count = $exons->count;
  my ($e_length, $e_gc);
  # Iterate over each exon
  while (my $exon = $exons->next) {
    $e_length += $exon->length;
    $e_gc     += $exon->gc_content;
  }
  print join "\t", int($e_length/$e_count), $e_gc/$e_count;
  print "\n";
  # Introns don't exist in the dataset, so GAL
  # will infer them on the fly.
  my $introns = $mrna->introns;
  my $i_count = $introns->count;
  my ($i_length, $i_gc);
  while (my $intron = $introns->next) {
    $i_length += $intron->length;
    $i_gc     += $intron->gc_content;
  }
  print $i_length ? join("\t", int($i_length/$i_count), '',
			 $i_gc/$i_count) : "\t\t";
  print "\n";
}