|
0
|
1 #!/usr/bin/perl -w
|
|
|
2
|
|
|
3 # Author: lh3
|
|
|
4
|
|
|
5 # This script calculates a score using the BLAST scoring
|
|
|
6 # system. However, I am not sure how to count gap opens and gap
|
|
|
7 # extensions. It seems to me that column 5-8 are not what I am
|
|
|
8 # after. This script counts gaps from the last three columns. It does
|
|
|
9 # not generate reference skip (N) in the CIGAR as it is not easy to
|
|
|
10 # directly tell which gaps correspond to introns.
|
|
|
11
|
|
|
12 use strict;
|
|
|
13 use warnings;
|
|
|
14 use Getopt::Std;
|
|
|
15
|
|
|
16 my %opts = (a=>1, b=>3, q=>5, r=>2);
|
|
|
17 getopts('a:b:q:r:', \%opts);
|
|
|
18 die("Usage: psl2sam.pl [-a $opts{a}] [-b $opts{b}] [-q $opts{q}] [-r $opts{r}] <in.psl>\n") if (@ARGV == 0 && -t STDIN);
|
|
|
19
|
|
|
20 my @stack;
|
|
|
21 my $last = '';
|
|
|
22 my ($a, $b, $q, $r) = ($opts{a}, $opts{b}, $opts{q}, $opts{r});
|
|
|
23 while (<>) {
|
|
|
24 next unless (/^\d/);
|
|
|
25 my @t = split;
|
|
|
26 my @s;
|
|
|
27 my $cigar = '';
|
|
|
28 if ($t[8] eq '-') {
|
|
|
29 my $tmp = $t[11];
|
|
|
30 $t[11] = $t[10] - $t[12];
|
|
|
31 $t[12] = $t[10] - $tmp;
|
|
|
32 }
|
|
|
33 @s[0..4] = ($t[9], (($t[8] eq '+')? 0 : 16), $t[13], $t[15]+1, 0);
|
|
|
34 @s[6..10] = ('*', 0, 0, '*', '*');
|
|
|
35 $cigar .= $t[11].'H' if ($t[11]); # 5'-end clipping
|
|
|
36 my @x = split(',', $t[18]);
|
|
|
37 my @y = split(',', $t[19]);
|
|
|
38 my @z = split(',', $t[20]);
|
|
|
39 my ($y0, $z0) = ($y[0], $z[0]);
|
|
|
40 my ($gap_open, $gap_ext) = (0, 0, 0);
|
|
|
41 for (1 .. $t[17]-1) {
|
|
|
42 my $ly = $y[$_] - $y[$_-1] - $x[$_-1];
|
|
|
43 my $lz = $z[$_] - $z[$_-1] - $x[$_-1];
|
|
|
44 if ($ly < $lz) { # del: the reference gap is longer
|
|
|
45 ++$gap_open;
|
|
|
46 $gap_ext += $lz - $ly;
|
|
|
47 $cigar .= ($y[$_] - $y0) . 'M';
|
|
|
48 $cigar .= ($lz - $ly) . 'D';
|
|
|
49 ($y0, $z0) = ($y[$_], $z[$_]);
|
|
|
50 } elsif ($lz < $ly) { # ins: the query gap is longer
|
|
|
51 ++$gap_open;
|
|
|
52 $gap_ext += $ly - $lz;
|
|
|
53 $cigar .= ($z[$_] - $z0) . 'M';
|
|
|
54 $cigar .= ($ly - $lz) . 'I';
|
|
|
55 ($y0, $z0) = ($y[$_], $z[$_]);
|
|
|
56 }
|
|
|
57 }
|
|
|
58 $cigar .= ($t[12] - $y0) . 'M';
|
|
|
59 $cigar .= ($t[10] - $t[12]).'H' if ($t[10] != $t[12]); # 3'-end clipping
|
|
|
60 $s[5] = $cigar;
|
|
|
61 my $score = $a * $t[0] - $b * $t[1] - $q * $gap_open - $r * $gap_ext;
|
|
|
62 $score = 0 if ($score < 0);
|
|
|
63 $s[11] = "AS:i:$score";
|
|
|
64 print join("\t", @s), "\n";
|
|
|
65 }
|
