annotate methylation_analysis_bismark/methylation_analysis/bismark_wrapper.pl @ 15:d2ba233899b0 draft default tip

Uploaded
author fcaramia
date Mon, 17 Jun 2013 21:50:56 -0400
parents d15b4a2e3bdc
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
1 use strict;
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
2 use warnings;
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
3 use File::Basename;
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
4 use Cwd;
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
5 use File::Path qw(make_path remove_tree);
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
6 die qq(
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
7 Bad numbr of inputs
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
8
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
9 ) if(!@ARGV);
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
10
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
11 my @bam_list_entries;
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
12
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
13 my $player_options = "--bowtie2 -p 4 ";
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
14 my $bam_output;
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
15 my $summary;
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
16 my $genome;
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
17 my $singles = "";
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
18 my $mates1 = "";
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
19 my $mates2 = "";
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
20 my $directional="";
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
21 my $format="";
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
22 my $log_file="";
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
23
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
24
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
25 foreach my $input (@ARGV)
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
26 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
27 my @tmp = split "::", $input;
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
28 if($tmp[0] eq "GENOME")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
29 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
30 $genome=$tmp[1];
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
31 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
32 elsif($tmp[0] eq "MATES")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
33 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
34 $mates1 = $tmp[1];
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
35 $mates2 = $tmp[2];
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
36 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
37 elsif($tmp[0] eq "SINGLES")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
38 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
39 $singles = $tmp[1];
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
40 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
41 elsif($tmp[0] eq "FORMAT")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
42 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
43 $format = $tmp[1];
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
44 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
45 elsif($tmp[0] eq "DIRECTIONAL")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
46 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
47 $directional = $tmp[1];
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
48 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
49 elsif($tmp[0] eq "SUMMARY")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
50 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
51 $summary = $tmp[1];
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
52 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
53 elsif($tmp[0] eq "OUTPUT")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
54 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
55 $bam_output = $tmp[1];
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
56 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
57
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
58 else
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
59 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
60 die("Unknown Input: $input\n");
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
61 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
62 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
63
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
64
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
65 my $working_dir = cwd()."/BISMARK_OUTPUT";
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
66 make_path($working_dir);
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
67
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
68
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
69 #run bismark
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
70
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
71 if ($singles eq "")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
72 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
73 system ("bismark $player_options $directional $format $genome -o $working_dir -1 $mates1 -2 $mates2 > $summary 2>&1 ");
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
74
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
75 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
76 else
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
77 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
78 system ("bismark $player_options $directional $format $genome -o $working_dir $singles > $summary 2>&1");
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
79
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
80 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
81
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
82
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
83 move_files($working_dir);
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
84
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
85
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
86 sub move_files
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
87 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
88 my $name;
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
89 my $suffix;
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
90 my $path;
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
91 my $local_dir = $_[0];
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
92 opendir(DIR, $local_dir);
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
93 #print ("Openning: $local_dir\n");
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
94 my @FILES= readdir(DIR);
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
95 closedir(DIR);
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
96 foreach my $file (@FILES)
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
97 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
98 if ($file eq "." || $file eq "..")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
99 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
100 #print ("./ or ../ skipped\n");
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
101 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
102 elsif (-d "$local_dir/$file")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
103 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
104 #print ("moving to: $local_dir/$file\n");
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
105 move_files("$local_dir/$file");
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
106 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
107 elsif (-f "$local_dir/$file")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
108 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
109 ($name,$path,$suffix) = fileparse($file,qr"\.[^.]*$");
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
110 if ($suffix eq ".sam")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
111 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
112 #converto to BAM and Delete SAM
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
113 system ("MergeSamFiles.sh MAX_RECORDS_IN_RAM=2000000 I=$local_dir/$file O=$bam_output VALIDATION_STRINGENCY=LENIENT SO=coordinate USE_THREADING=true CREATE_INDEX=true > /dev/null 2>&1");
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
114 system ("rm -rf $local_dir/$file");
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
115
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
116 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
117 elsif($suffix eq ".txt")
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
118 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
119 system ("mv $local_dir/$file $summary");
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
120 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
121 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
122 else
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
123 {
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
124 die("Unrecognized file generated: $file\n");
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
125 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
126
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
127 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
128 }
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
129
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
130
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
131
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
132
d15b4a2e3bdc Uploaded
fcaramia
parents:
diff changeset
133