vcfindelproximity

#!/usr/bin/env perl
# Show SNPs around an INDEL



# for line in the vcf
# stuff the line into a queue
# when you reach an indel
# record the position
# pop lines from the back of the queue until we are at the current position
#

my @lines;

my $prox = $ARGV[0];

my $lastchrom = "";
my $indelpos = 0;

while (<STDIN>) {

    if ($_ =~ /^#/) {
        print $_;
        next;
    }

    $_ =~ /^(.+?)\t(.+?)\t(.+?)\t(.+?)\t(.+?)\t/;
    my $chrom = $1;
    my $pos = $2;
    my $tag = $3;
    my $ref = $4;
    my $alt = $5;
    #print "chrom: $chrom, pos: $pos, ref: $ref, alt: $alt\n";

    # if new chrom, print out everything from last one
    if ($lastchrom == "") {
        $lastchrom = $chrom;
    }

    if ($chrom != $lastchrom) {
        while ($lines) {
            print pop(@lines);
        }
    }

    unshift(@lines, $_);

    my $diff = length($ref) - length($alt);

    if ($diff != 0) {
        # insertion
        if ($indelpos == 0) {
            $indelpos = $pos;
        }
        $nextindelpos = $pos;
        #print "last $indelpos next $nextindelpos\n";
        while (@lines) {
            my $line = pop(@lines);
            $line =~ /^(.+?)\t(.+?)\t(.+?)\t(.+?)\t(.+?)\t/;
            my $c = $1;
            my $p = $2;
            my $t = $3;
            my $r = $4;
            my $a = $5;
            # print indels
            if (length($r) - length($a) != 0) {
                print $line;
            } else {
            # print other events which are more than prox away from indels
                if (abs($indelpos - $p) >= $prox and abs($nextindelpos - $p) >= $prox) {
                    print $line;
                }
            }
        }
        $indelpos = $pos;
    }
}

# flush lines end of file
while ($lines) {
    print pop(@lines);
}