|
0
|
1 #!/usr/bin/perl -w
|
|
|
2 use strict;
|
|
|
3
|
|
|
4 #divide the alleles and their information into separate columns for pgSnp-like
|
|
|
5 #files. Keep any additional columns beyond the pgSnp ones.
|
|
|
6 #reads from stdin, writes to stdout
|
|
|
7 my $ref;
|
|
|
8 my $in;
|
|
|
9 if (@ARGV && $ARGV[0] =~ /-ref=(\d+)/) {
|
|
|
10 $ref = $1 -1;
|
|
|
11 if ($ref == -1) { undef $ref; }
|
|
|
12 shift @ARGV;
|
|
|
13 }
|
|
|
14 if (@ARGV) {
|
|
|
15 $in = shift @ARGV;
|
|
|
16 }
|
|
|
17
|
|
|
18 open(FH, $in) or die "Couldn't open $in, $!\n";
|
|
|
19 while (<FH>) {
|
|
|
20 chomp;
|
|
|
21 my @f = split(/\t/);
|
|
|
22 if ($f[0] =~ /^\d+$/ && $f[1] =~ /chr/) { #has bin column shift list
|
|
|
23 shift @f; #remove bin
|
|
|
24 }
|
|
|
25 my @a = split(/\//, $f[3]);
|
|
|
26 my @fr = split(/,/, $f[5]);
|
|
|
27 my @sc = split(/,/, $f[6]);
|
|
|
28 if ($f[4] == 1) { #homozygous add N, 0, 0
|
|
|
29 if ($ref) { push(@a, $f[$ref]); }
|
|
|
30 else { push(@a, "N"); }
|
|
|
31 push(@fr, 0);
|
|
|
32 push(@sc, 0);
|
|
|
33 }
|
|
|
34 if ($f[4] > 2) { next; } #skip those with more than 2 alleles
|
|
|
35 print "$f[0]\t$f[1]\t$f[2]\t$a[0]\t$fr[0]\t$sc[0]\t$a[1]\t$fr[1]\t$sc[1]";
|
|
|
36 if (scalar @f > 7) {
|
|
|
37 splice(@f, 0, 7); #remove first 7
|
|
|
38 print "\t", join("\t", @f), "\n";
|
|
|
39 }else { print "\n"; }
|
|
|
40 }
|
|
|
41 close FH;
|
|
|
42
|
|
|
43 exit;
|
|
|
44
|
