extract_run_tallies

#!/usr/bin/env perl

use strict;
use Getopt::Long;

my $print_usage = 0;

my $usage = <<USAGE;

  This script tabulates Illumina flow cell serial numbers from mapped reads
  in a SAM file

  Reads from stdin writes to stdout

  Mark Stenglein,  10/5/2023

  Usage: $0 [-h] [-r regex]

   [-h]          print this message

   [-r regex]    optional regex to search for in refseq names

USAGE

if ((scalar @ARGV == 0) and -t STDIN) { print $usage and exit; }

my $regex = undef;

GetOptions ("h" => \$print_usage,
            "r=s" => \$regex);

my %run_tally = ();

while (<>)
{
   chomp;
   my @fields = split "\t";
   my $read = $fields[0];
   my @read_parts = split ":", $read;
   my $run = $read_parts[2];
   my $refseq = $fields[2];

   if (defined $regex and ($refseq =~ /$regex/))
   {
      $run_tally{$run} += 1;
   }
   elsif (!defined $regex)
   {
      $run_tally{$run} += 1;
   }
}

foreach my $run (keys %run_tally)
{
   print "$run\t$run_tally{$run}\n";
}