annotate PsiCLASS-1.0.2/samtools-0.1.19/misc/bowtie2sam.pl @ 0:903fc43d6227 draft default tip

Uploaded
author lsong10
date Fri, 26 Mar 2021 16:52:45 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
1 #!/usr/bin/perl -w
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
2
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
3 # Contact: lh3
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
4 # Version: 0.1.1
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
5
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
6 use strict;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
7 use warnings;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
8 use Getopt::Std;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
9
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
10 &bowtie2sam;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
11 exit;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
12
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
13 sub bowtie2sam {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
14 my %opts = ();
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
15 die("Usage: bowtie2sam.pl <aln.bowtie>\n") if (@ARGV == 0 && -t STDIN);
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
16 # core loop
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
17 my (@s, $last, @staging, $k, $best_s, $subbest_s, $best_k);
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
18 $last = '';
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
19 while (<>) {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
20 my ($name, $nm) = &bowtie2sam_aux($_, \@s); # read_name, number of mismatches
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
21 if ($name eq $last) {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
22 # I do not know whether the multiple hits are ordered on the
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
23 # number of mismatches. I assume they are not and so I have to
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
24 # keep all these multiple hits in memory.
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
25 @{$staging[$k]} = @s;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
26 if ($best_s > $nm) {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
27 $subbest_s = $best_s;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
28 $best_s = $nm;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
29 $best_k = $k;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
30 } elsif ($subbest_s > $nm) {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
31 $subbest_s = $nm;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
32 }
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
33 ++$k;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
34 } else {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
35 if ($last) {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
36 if ($best_s == $subbest_s) {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
37 $staging[$best_k][4] = 0;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
38 } elsif ($subbest_s - $best_s == 1) {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
39 $staging[$best_k][4] = 15 if ($staging[$best_k][4] > 15);
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
40 }
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
41 print join("\t", @{$staging[$best_k]}), "\n";
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
42 }
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
43 $k = 1; $best_s = $nm; $subbest_s = 1000; $best_k = 0;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
44 @{$staging[0]} = @s;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
45 $last = $name;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
46 }
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
47 }
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
48 print join("\t", @{$staging[$best_k]}), "\n" if ($best_k >= 0);
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
49 }
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
50
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
51 sub bowtie2sam_aux {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
52 my ($line, $s) = @_;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
53 chomp($line);
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
54 my @t = split("\t", $line);
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
55 my $ret;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
56 @$s = ();
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
57 # read name
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
58 $s->[0] = $ret = $t[0];
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
59 $s->[0] =~ s/\/[12]$//g;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
60 # initial flag (will be updated later)
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
61 $s->[1] = 0;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
62 # read & quality
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
63 $s->[9] = $t[4]; $s->[10] = $t[5];
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
64 # cigar
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
65 $s->[5] = length($s->[9]) . "M";
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
66 # coor
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
67 $s->[2] = $t[2]; $s->[3] = $t[3] + 1;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
68 $s->[1] |= 0x10 if ($t[1] eq '-');
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
69 # mapQ
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
70 $s->[4] = $t[6] == 0? 25 : 0;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
71 # mate coordinate
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
72 $s->[6] = '*'; $s->[7] = $s->[8] = 0;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
73 # aux
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
74 my $nm = @t - 7;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
75 push(@$s, "NM:i:" . (@t-7));
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
76 push(@$s, "X$nm:i:" . ($t[6]+1));
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
77 my $md = '';
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
78 if ($t[7]) {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
79 $_ = $t[7];
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
80 my $a = 0;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
81 while (/(\d+):[ACGTN]>([ACGTN])/gi) {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
82 my ($y, $z) = ($1, $2);
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
83 $md .= (int($y)-$a) . $z;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
84 $a += $y - $a + 1;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
85 }
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
86 $md .= length($s->[9]) - $a;
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
87 } else {
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
88 $md = length($s->[9]);
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
89 }
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
90 push(@$s, "MD:Z:$md");
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
91 return ($ret, $nm);
903fc43d6227 Uploaded
lsong10
parents:
diff changeset
92 }