annotate createControlSubSet.pl @ 2:106d0a5c2b7f draft

Uploaded
author jbrayet
date Tue, 02 Feb 2016 09:02:40 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
1 #!/usr/bin/perl
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
2
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
3 #filter out dulpicates from SAMPLE (optional) and create a control dataset w/o duplicates with the same number of reads as in the SAMPLE
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
4
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
5 use strict;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
6 use warnings;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
7 use diagnostics;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
8
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
9 my $usage = qq{
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
10 $0
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
11
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
12 -----------------------------
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
13 mandatory parameters:
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
14
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
15 -f CHiP_file
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
16 -c control_file
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
17 -t type [bam, sam, eland]
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
18 -o output file
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
19 -----------------------------
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
20 optional parameters:
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
21
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
22 none
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
23 };
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
24
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
25 if(scalar(@ARGV) == 0){
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
26 print $usage;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
27 exit(0);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
28 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
29
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
30 ## mandatory arguments
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
31
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
32 my $filename = "";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
33 my $output_fname = "";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
34
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
35 my $controlFilename = "";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
36 my $type = "";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
37 my $sampleOutput = "";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
38
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
39
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
40 ## optional arguments
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
41
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
42 ## parse command line arguments
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
43
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
44 while(scalar(@ARGV) > 0){
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
45 my $this_arg = shift @ARGV;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
46 if ( $this_arg eq '-h') {print "$usage\n"; exit; }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
47
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
48 elsif ( $this_arg eq '-f') {$filename = shift @ARGV;}
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
49 elsif ( $this_arg eq '-c') {$controlFilename = shift @ARGV;}
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
50 elsif ( $this_arg eq '-t') {$type = shift @ARGV;}
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
51 elsif ( $this_arg eq '-o') {$output_fname = shift @ARGV;}
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
52 elsif ( $this_arg eq '-s') {$sampleOutput = shift @ARGV;}
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
53
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
54
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
55 elsif ( $this_arg =~ m/^-/ ) { print "unknown flag: $this_arg\n";}
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
56 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
57
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
58 if ( $filename eq ""){
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
59 die "you should specify chip file\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
60 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
61 if( $controlFilename eq ""){
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
62 die "you should specify control file\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
63 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
64 if( $type eq ""){
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
65 die "you should specify file type (bam, sam or eland)\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
66 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
67 if( $output_fname eq ""){
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
68 die "you should specify output filename\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
69 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
70
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
71
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
72 print "\n-----------------\n\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
73
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
74 my %hash;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
75 my $chipCount = 0;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
76 my @header;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
77
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
78
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
79 if ($type eq "eland") {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
80 open FILE, "< $filename " || die "$filename : $!\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
81 while(<FILE>){
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
82 my @fields = split(/\t/,$_);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
83 my $entry = $fields[6].":".$fields[7]."-".$fields[8];
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
84 unless (exists($hash{$entry})) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
85 $hash{$entry} = $_;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
86 $chipCount++;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
87 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
88 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
89 } elsif ($type eq "sam") {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
90 open FILE, "< $filename " || die "$filename : $!\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
91 while(<FILE>){
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
92 if (m/^@/) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
93 push(@header,$_);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
94 next;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
95 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
96 my @fields = split(/\t/,$_);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
97 next if (scalar(@fields)<10);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
98 my $entry = $fields[2].":".$fields[3]."-".$fields[1];
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
99 unless (exists($hash{$entry})) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
100 $hash{$entry} = $_;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
101 $chipCount++;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
102 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
103 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
104 } elsif ($type eq "bam") {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
105 open(FILE, "samtools view -h $filename |") or die "$0: can't open ".$filename.":$!\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
106 while(<FILE>){
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
107 if (m/^@/) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
108 push(@header,$_);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
109 next;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
110 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
111 my @fields = split(/\t/,$_);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
112 next if (scalar(@fields)<10);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
113 my $entry = $fields[2].":".$fields[3]."-".$fields[1];
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
114 unless (exists($hash{$entry})) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
115 $hash{$entry} = $_;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
116 $chipCount++;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
117 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
118 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
119 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
120 close FILE;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
121 print "ChIP: $chipCount\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
122
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
123 if ($sampleOutput ne "") {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
124
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
125 open OUT, "> $sampleOutput" || die "$sampleOutput: $!\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
126
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
127 if ($type eq "bam" || $type eq "sam") { #print header
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
128 for my $headerLine (@header) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
129 print OUT $headerLine;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
130 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
131 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
132 for my $line (values %hash) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
133 print OUT $line;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
134 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
135 close OUT;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
136 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
137
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
138 delete @hash{keys %hash};
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
139 @header = ();
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
140
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
141 my $controlCount = 0;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
142 if ($type eq "eland") {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
143 open FILE, "< $controlFilename " || die "$controlFilename : $!\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
144 while(<FILE>){
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
145 my @fields = split(/\t/,$_);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
146 my $entry = $fields[6].":".$fields[7]."-".$fields[8];
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
147 unless (exists($hash{$entry})) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
148 $hash{$entry} = $_;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
149 $controlCount++;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
150 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
151 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
152 } elsif ($type eq "sam") {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
153 open FILE, "< $controlFilename " || die "$controlFilename : $!\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
154 while(<FILE>){
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
155 if (m/^@/) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
156 push(@header,$_);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
157 next;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
158 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
159 my @fields = split(/\t/,$_);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
160 my $entry = $fields[2].":".$fields[3]."-".$fields[1];
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
161 unless (exists($hash{$entry})) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
162 $hash{$entry} = $_;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
163 $controlCount++;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
164 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
165 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
166 } elsif ($type eq "bam") {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
167 open(FILE, "samtools view -h $controlFilename |") or die "$0: can't open ".$controlFilename.":$!\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
168 while(<FILE>){
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
169 if (m/^@/) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
170 push(@header,$_);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
171 next;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
172 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
173 my @fields = split(/\t/,$_);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
174 my $entry = $fields[2].":".$fields[3]."-".$fields[1];
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
175 unless (exists($hash{$entry})) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
176 $hash{$entry} = $_;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
177 $controlCount++;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
178 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
179 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
180 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
181 close FILE;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
182 print "Control: $controlCount\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
183 my $prob = $chipCount/$controlCount;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
184
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
185 open OUT, "> $output_fname" || die "$output_fname: $!\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
186
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
187 if ($type eq "bam" || $type eq "sam") { #print header
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
188 for my $headerLine (@header) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
189 print OUT $headerLine;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
190 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
191 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
192 my $count = 0;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
193
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
194 for my $line (values %hash) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
195 my $rand = rand();
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
196
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
197 if ($rand < $prob) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
198 print OUT $line;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
199 $count ++;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
200 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
201 last if ($count == $chipCount);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
202 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
203
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
204
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
205 if ($count < $chipCount) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
206
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
207 $prob = ($chipCount-$count)/$controlCount*1.1;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
208
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
209 for my $line (values %hash) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
210 my $rand = rand();
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
211
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
212 if ($rand < $prob) {
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
213 print OUT $line;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
214 $count ++;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
215 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
216 last if ($count == $chipCount);
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
217 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
218 }
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
219
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
220 print "count = $count\n";
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
221 close OUT;
106d0a5c2b7f Uploaded
jbrayet
parents:
diff changeset
222