annotate scripts/mergeTagsWithoutGap.pl @ 7:e96a0d0f8dd6 draft

planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
author portiahollyoak
date Mon, 25 Apr 2016 14:20:48 -0400
parents 28d1a6f8143f
children
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 #chr2L 735929 736005 HWUSI-EAS1533_0002:1:73:4665:12371#0/2 FBgn0000155_roo,-58;FBgn0000155_roo,-8722; -
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
3 use warnings;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
4 use strict;
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 if(scalar(@ARGV)<1 || grep {/^-h/} @ARGV)
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
7 {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
8 die "
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
9 usage: mergeOverlapBed4.pl inputFile
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
10 Expects BED input with at least 4 fields. For each {chr,name} pair,
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
11 merges overlapping ranges and prints out sorted BED4 to stdout.
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
12 inputFile can be - or stdin to read from stdin.
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
13 ";
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
14 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
15
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
16 my $input=shift @ARGV;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
17 grep {s/^stdin$/-/i} $input;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
18
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
19 my %item2coords;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
20 open IN,$input;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
21 while (<IN>)
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
22 {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
23 chomp;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
24 my ($chrom,$start,$end,$sample,$class,$strand)=split/\t/;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
25 die "Sorry, input must have at least 4 fields of BED.\n" if ! $class;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
26 # random choose one
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
27 # my @loc=$class=~/(.*?),(\+|-)(.*)/;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
28 # my $transposonStrand=($strand eq $loc[1])?"antisense":"sense";
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
29 # push @{$item2coords{"$chrom;$strand;$loc[0];$transposonStrand"}},[$start,$end,$sample]
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
30
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
31 # norm by class
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
32 my @loc=map { [/(.*?),(\+|-)(.*)/] } split/;/,$class;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
33 my %transposonName;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
34 foreach my $l (@loc)
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 my $transposonStrand=($strand eq $$l[1])?"antisense":"sense";
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
37 $transposonName{$$l[0]}=$transposonStrand;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
38 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
39 my $c=1/scalar(keys %transposonName);
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
40 push @{$item2coords{"$chrom;$strand;$_;$transposonName{$_}"}},[$start,$end,$sample,$c] foreach keys %transposonName;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
41 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
42 close IN;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
43
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
44 my @results;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
45 foreach my $item (keys %item2coords)
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 my @sortedCoords=sort{ $a->[0]<=>$b->[0] } @{$item2coords{$item}};
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
48 my ($chrom,$strand,$tName,$tStrand)=split(/;/,$item);
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
49 my ($mergeStart,$mergeEnd,$mergeSample,$mergeCounts)=@{shift @sortedCoords};
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
50 my %sampleCounts;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
51 $sampleCounts{$mergeSample}=$mergeCounts;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
52 foreach my $rangeRef (@sortedCoords)
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
53 {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
54 my ($rangeStart,$rangeEnd,$rangeSample,$rangeCounts)=@{$rangeRef};
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
55 if($rangeEnd<=$mergeEnd)
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
56 {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
57 $sampleCounts{$rangeSample}+=$rangeCounts;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
58 next;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
59 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
60 if($rangeStart>=$mergeEnd)
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
61 {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
62 my $count="";
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
63 $count.=$_.",".$sampleCounts{$_}.";" foreach keys %sampleCounts;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
64 push @results,[$chrom,$mergeStart,$mergeEnd,$tName,$count,$strand,$tStrand];
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
65 ($mergeStart,$mergeEnd,$mergeSample,$mergeCounts)=($rangeStart,$rangeEnd,$rangeSample,$rangeCounts);
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
66 %sampleCounts=();
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
67 $sampleCounts{$mergeSample}=$mergeCounts;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
68 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
69 else
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
70 {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
71 $mergeEnd=$rangeEnd;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
72 $sampleCounts{$rangeSample}+=$rangeCounts;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
73 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
74 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
75 my $count="";
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
76 $count.=$_.",".$sampleCounts{$_}.";" foreach keys %sampleCounts;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
77 push @results,[$chrom,$mergeStart,$mergeEnd,$tName,$count,$strand,$tStrand] if $mergeEnd;
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
78 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
79
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
80 sub bed4Cmp
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
81 {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
82 # For sorting by chrom, chromStart, and names -- reverse order for names
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
83 return $a->[0] cmp $b->[0] ||
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
84 $a->[1] <=> $b->[1] ||
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
85 $b->[3] cmp $a->[3];
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
86 }
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
87
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
88 foreach my $r (sort bed4Cmp @results)
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
89 {
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
90 print join("\t",@{$r}),"\n";
28d1a6f8143f planemo upload for repository https://github.com/portiahollyoak/Tools commit 132bb96bba8e7aed66a102ed93b7744f36d10d37-dirty
portiahollyoak
parents:
diff changeset
91 }