annotate scripts/pickUniqPairFastq_MEM.pl @ 19:e541089245ba draft

planemo upload for repository https://github.com/portiahollyoak/Tools commit 264ceb3f8eae0938d6c97d0e01985cd9ce511467
author mvdbeek
date Thu, 01 Sep 2016 11:11:59 -0400
parents ca36262102d8
children 9672fe07a232
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
1 #!/share/bin/perl
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
2 use Bio::Seq;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
3
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
4 die "perl $0 <sam> <output prefix>\n" if @ARGV<1;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
5
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
6 open m1,">$ARGV[1].1.fastq";
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
7 open m2,">$ARGV[1].2.fastq";
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
8
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
9 open in,$ARGV[0];
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
10 my %pe;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
11 while(<in>)
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
12 {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
13 chomp;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
14 my @f=split/\t/,$_,12;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
15 ## read number 1 or 2
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
16 my ($rnum)=$f[1]=~/(\d)$/;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
17
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
18 ## XT:A:*
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
19 my $xt="";
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
20 my @a=split(/\s+/, $_);
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
21 my $as=0;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
22 my $xs=0;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
23 for my $i (11..$#a) {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
24 if ($a[$i] =~ /^AS:i:/) {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
25 $a[$i] =~ s/AS:i://;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
26 $as=$a[$i];
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
27 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
28 elsif ($a[$i] =~ /^XS:i:/) {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
29 $a[$i] =~ s/XS:i://;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
30 $xs=$a[$i];
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
31 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
32 if (($xs > 0) && ($as-$xs <= $ARGV[2])) {$xt="R";}
12
ca36262102d8 planemo upload for repository https://github.com/portiahollyoak/Tools commit 5d021f520b653582862ec98dd812a051b804aa50
portiahollyoak
parents: 0
diff changeset
33 elsif ($as > 0) {$xt="U";}
0
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
34 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
35
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
36 ## revcom the read mapped to the reverse strand
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
37 if($f[1]=~/r/)
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
38 {
12
ca36262102d8 planemo upload for repository https://github.com/portiahollyoak/Tools commit 5d021f520b653582862ec98dd812a051b804aa50
portiahollyoak
parents: 0
diff changeset
39 my $seq=Bio::Seq->new(-seq=>$f[9], -alphabet => 'dna');
0
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
40 $f[9]=$seq->revcom->seq;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
41 $f[10]=reverse $f[10];
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
42 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
43 if (($rnum == 1) || ($rnum == 2))
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
44 {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
45 ${$pe{$f[0]}}[$rnum-1]=[$xt,$f[9],$f[10]];
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
46 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
47 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
48 close in;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
49
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
50 foreach my $id (keys %pe)
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
51 {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
52 my @rid=@{$pe{$id}};
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
53 if (($rid[0][1] ne "") && ($rid[1][1] ne "") && (($rid[0][0] eq "U" || $rid[1][0] eq "U")))
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
54 {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
55 print m2 "@"."$id/2","\n",$rid[1][1],"\n","+$id/2","\n",$rid[1][2],"\n";
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
56 print m1 "@"."$id/1","\n",$rid[0][1],"\n","+$id/1","\n",$rid[0][2],"\n";
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
57 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
58 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
59 close m1;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
60 close m2;