mjclark

#This is probably not the best way to go about this, but it basically works.
#Keep in mind that PERL's "rand()" function isn't really random, but for this purpose it doesn't really matter.
#Suggested usage: samtools view -h <in.bam> | perl random_line_extraction.using_ratio.pl <ratio in decimal form> | samtools view -bS - > <out.bam>

$num=shift;
while($line=<STDIN>) {
    chomp($line);
    if($line =~ m/^@/) {
        print "$line\n";
    }

mjclark

#Command: cat in.bed | perl windowize-bed.pl <windowsize> > out.bed
$windowsize=shift;
$ctr=0;
print STDERR "$ctr";
while($line=<STDIN>) {
    $ctr++;
    print STDERR "\r$ctr";
    chomp($line);
    @line=split(/\t/, $line);
    for($i=$line[1]; $i<$line[2]-$windowsize+1; $i+=$windowsize) {

mjclark

#Automatically calculates mean heterozygous allele balance for heterozygous alleles in VCF4 files
#Allele balance must be pre-caluclated and defined for each variant in advance and given the INFO ID flag "AB" (default for GATK, etc.)

usage="usage: sh mean_allele_balance.sh <input.vcf>";
if (( $# != 1 )); then echo $usage; exit; fi

#input
VCF=$1;

#calculations