annotate vcftools_main/VCFToolFilter/VCFToolsFilter.pl @ 1:0f67ed444d47 draft

Uploaded
author gandres
date Thu, 02 Jul 2015 11:07:45 -0400
parents 3b1436a9a6e5
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
1
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
2 #!/usr/bin/perl
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
3
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
4 use strict;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
5 use Switch;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
6 use Getopt::Long;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
7 use Bio::SeqIO;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
8
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
9 my $usage = qq~Usage:$0 <args> [<opts>]
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
10
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
11 where <args> are:
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
12
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
13 -i, --input <VCF input>
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
14 -o, --out <Output basename>
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
15
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
16 <opts> are:
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
17
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
18 -s, --samples <Samples to be analyzed. Comma separated list>
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
19 -c, --chromosomes <Chromosomes to be analyzed. Comma separated list>
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
20 -e, --export <Output format (VCF/freq/plink. Default: VCF>
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
21 -f, --frequency <Minimum MAF. Default: 0.001>
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
22 -m, --max_freq <Maximum MAF. Default: 0.5>
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
23 -a, --allow_missing <Allowed missing data proportion per site. Must be comprised between 0 and 1. Default: 1>
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
24 -n, --nb_alleles <Accepted number of alleles (min,max). Default: 2,4>
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
25 -t, --type <Type of polymorphisms to keep (ALL/SNP/INDEL). Default: ALL>
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
26 -b, --bounds <Lower bound and upper bound for a range of sites to be processed (start,end). Default: 1, 100000000>
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
27 ~;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
28 $usage .= "\n";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
29
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
30 my ($input,$out);
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
31
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
32
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
33 #my $indel_size_max = 500;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
34 #my $indel_size_min = 1;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
35 my $frequency_max = 0.5;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
36 my $frequency_min = 0.001;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
37 my $pos_max = 100000000000;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
38 my $pos_min = 0;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
39 my $filter_snp_type = "all";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
40
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
41 my $missing_data = 1;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
42 my $export = "VCF";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
43 my $type = "ALL";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
44 my $nb_alleles;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
45 my $bounds;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
46 my $samples;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
47 my $chromosomes;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
48
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
49 GetOptions(
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
50 "input=s" => \$input,
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
51 "out=s" => \$out,
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
52 "samples=s" => \$samples,
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
53 "chromosomes=s" => \$chromosomes,
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
54 "frequency=s" => \$frequency_min,
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
55 "max_freq=s" => \$frequency_max,
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
56 "allow_missing=s"=> \$missing_data,
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
57 "export=s" => \$export,
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
58 "type=s" => \$type,
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
59 "nb_alleles=s" => \$nb_alleles,
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
60 "bounds=s" => \$bounds,
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
61 );
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
62
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
63
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
64 die $usage
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
65 if ( !$input || !$out);
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
66
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
67
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
68 my @dnasamples;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
69 if ($samples)
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
70 {
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
71 @dnasamples = split(",",$samples);
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
72 }
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
73 my @nalleles;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
74 if ($nb_alleles)
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
75 {
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
76 @nalleles = split(",",$nb_alleles);
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
77 }
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
78 my @boundaries;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
79 if ($bounds)
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
80 {
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
81 @boundaries = split(",",$bounds);
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
82 }
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
83 my @chromosomes_list;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
84 if ($chromosomes)
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
85 {
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
86 @chromosomes_list = split(",",$chromosomes);
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
87 }
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
88
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
89
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
90 my $experiment = "chromosomes";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
91 my $table = "";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
92 my %genes;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
93 my @snp_ids;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
94 my @snp_ids_and_positions;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
95 my @snp_ids_and_positions_all;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
96 my $gene;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
97 my $snp_num = 0;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
98 my %ref_sequences;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
99 my %snps_of_gene;
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
100
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
101
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
102
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
103
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
104 my $indiv_cmd = "";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
105 if (@dnasamples)
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
106 {
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
107 $indiv_cmd = "--indv " . join(" --indv ",@dnasamples);
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
108 }
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
109
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
110 my $chrom_cmd = "";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
111 if (@chromosomes_list)
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
112 {
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
113 $chrom_cmd = "--chr " . join(" --chr ",@chromosomes_list);
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
114 }
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
115
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
116 my $export_cmd = "--recode";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
117 if ($export eq "freq")
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
118 {
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
119 $export_cmd = "--freq";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
120 }
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
121 if ($export eq "plink")
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
122 {
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
123 $export_cmd = "--plink";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
124 }
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
125
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
126
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
127
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
128 my $nb_alleles_cmd = "--min-alleles 1 --max-alleles 4";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
129 if (@nalleles)
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
130 {
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
131 $nb_alleles_cmd = "--min-alleles $nalleles[0] --max-alleles $nalleles[1]";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
132 }
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
133 my $bounds_cmd = "--from-bp 1 --to-bp 100000000";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
134 if (@boundaries)
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
135 {
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
136 $bounds_cmd = "--from-bp $boundaries[0] --to-bp $boundaries[1]";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
137 }
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
138
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
139
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
140 my $type_cmd = "";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
141 if ($type eq "INDEL")
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
142 {
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
143 $type_cmd = "--keep-only-indels";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
144 }
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
145 if ($type eq "SNP")
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
146 {
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
147 $type_cmd = "--remove-indels";
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
148 }
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
149
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
150
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
151 system("vcftools --vcf $input --out $out --keep-INFO-all --remove-filtered-all $type_cmd $export_cmd $chrom_cmd $indiv_cmd $nb_alleles_cmd --maf $frequency_min --max-maf $frequency_max --max-missing $missing_data >>vcftools.log 2>&1");
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
152
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
153
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
154
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
155
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
156
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
157
3b1436a9a6e5 Uploaded
gandres
parents:
diff changeset
158