Non-ref      "A"     "A1"    "A2" ....... (column6)
A               AT      5       15            INTRONIC

 
#!/usr/bin/perl
use strict;
use warnings;
#initialize data
my(@data,$row,@dataline,$dnafile);
#open file to be read or quit
open(FILE1,"<",'file.txt')or die $!;
#initialize counts of variables
my $countExonic=0;
my $countIntergenic=0;
my $countIntronic=0;
my $countUpstream=0;
my $countDownstream=0;
my $countUTR5=0;
my $countUTR3=0;
my $countFrameshift=0;
my $countNonsynonymous=0;
my $countSyn=0;
my $countStopgain=0;
my $countStoploss=0;
my $countSplicing=0;
my $countErrors=0;
#set data equal to the previously opened file
@data=<FILE1>;close(FILE1);
#loop to grep rows of file that have genotype that matches non-ref allele
#$dataline[1] is genotype, $dataline[0] is non-ref allele, $dataline[7] is
#mutation type
#selects for ratio above .5
#the number for each type of mutation occurring in the grepped data is recorded
foreach $row(@data) {
        next unless $row=~/^\d/;
        @dataline=split(/\s+/,$row);
        next unless $dataline[1] ne "NA";
        next unless $dataline[1]=~/$dataline[0]/i;
        next unless $dataline[2]>=10;
        next unless $dataline[2]/$dataline[3]>=.5;
        if($dataline[7] =~ /exonic/){
                ++$countExonic;
        }elsif($dataline[7]=~/inter/){
                ++$countIntergenic;
        }elsif($dataline[7]=~/intron/){
  ++$countIntronic;
        }elsif($dataline[7]=~/upstream/){
                ++$countUpstream;
        }elsif($dataline[7]=~/downstream/){
                ++$countDownstream;
        }elsif($dataline[7]=~/UTR5/){
                ++$countUTR5;
        }elsif($dataline[7]=~/UTR3/){
                ++$countUTR3;
        }elsif($dataline[7]=~/frame/){
                ++$countFrameshift;
        }elsif($dataline[7]=~/nonsyn/){
                ++$countNonsynonymous;
        }elsif($dataline[7]=~/syn/){
                ++$countSyn;
        }elsif($dataline[7]=~/gain/){
                ++$countStopgain;
        }elsif($dataline[7]=~/loss/){
                ++$countStoploss;
        }elsif($dataline[7]=~/splic/){
                ++$countSplicing;
        }else{++$countErrors;
        }
}
#print out results of mutation counts
print "nc_RNA Exonic Mutations=$countExonic\n\n";
print "Intergenic Mutations=$countIntergenic\n\n";
print "Intronic Mutations=$countIntronic\n\n";
print "Upstream Mutations=$countUpstream\n\n";
print "Downstream Mutations=$countDownstream\n\n";
print "UTR5 Mutations=$countUTR5\n\n";
print "UTR3 Mutations=$countUTR3\n\n";
print "Frameshift Mutations=$countFrameshift\n\n";
print "Nonsynonymous Mutations=$countNonsynonymous\n\n";
print "Synonymous Mutations=$countSyn\n\n";
print "Stop Gain Mutations=$countStopgain\n\n";
print "Stop Loss Mutations=$countStoploss\n\n";
print "Splicing Mutations=$countSplicing\n\n";
print "Others=$countErrors\n\n";
exit;

 
Non-ref           "A"      "A1"       "A2"          "B"          "B1"         "B2"      
A                   AT       5            15          AA           13           14

 
next unless $dataline[1] ne "NA";
next unless $dataline[1]=~/$dataline[0]/i;
next unless $dataline[2]>=10;
next unless $dataline[2]/$dataline[3]>=.5;