annotate CreateMatrix.pl @ 3:0354759bf0ce draft default tip

Uploaded
author jnavarro
date Thu, 26 Oct 2017 07:37:25 -0400
parents e94977a3dcad
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
1 #!/usr/bin/perl
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
2 #V1.0.0
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
3 use strict;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
4 use warnings;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
5 use Getopt::Long;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
6
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
7 my $input_pileup_file;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
8 my $input_mpileupFiltred_file;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
9 my $input_name;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
10
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
11 GetOptions (
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
12 "input_name=s" => \$input_name,
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
13 "input_pileup_file=s" => \$input_pileup_file,
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
14 "input_mpileupFiltred_file=s" => \$input_mpileupFiltred_file
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
15 ) or die("Error in command line arguments\n");
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
16
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
17
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
18 print "Chrom\tPos\tRef\t$input_name\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
19
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
20 my %mpileupFiltred;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
21 open (VU,$input_mpileupFiltred_file) or die ("Can't open mpileupFiltred file\n");
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
22 while (my $line=<VU>){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
23 if ($line!~/^\s*$/){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
24 my @fields = split (/\t+/,$line);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
25 my $chr;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
26 my $pos;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
27 if ($fields[0]=~/^\s*([\w\-]+)\s*$/){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
28 $chr = $1;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
29 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
30 else {
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
31 print STDERR "Unable to detect chromosome in pileup file\n$line",$fields[0],"\n",$fields[1],"\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
32 exit(0);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
33 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
34 if ($fields[1]=~/^\s*(\d+)\s*$/){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
35 $pos = $1;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
36 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
37 else {
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
38 print STDERR "Unable to detect position in pileup file\n$line",$fields[0],"\n",$fields[1],"\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
39 exit(0);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
40 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
41 if ($#fields>=4){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
42 $mpileupFiltred{"$chr#$pos"}=$fields[4];
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
43 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
44 else {
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
45 print STDERR "Unrecognized pileup format, need at least 5 columns\n$line\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
46 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
47 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
48 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
49 close VU;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
50
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
51 my %covered;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
52
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
53 my $compt=0;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
54 open (PU,$input_pileup_file) or die ("Can't open pileup file\n");
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
55 while (my $line=<PU>){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
56 #$compt++;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
57 #if ($compt>200000){last;}
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
58 if ($line!~/^\s*$/){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
59 my @fields = split (/\t+/,$line);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
60 my $chr;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
61 my $pos;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
62 my $ref;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
63 my $depth;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
64 my $pile;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
65
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
66 if ($#fields>=4){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
67 if ($fields[0]=~/^\s*([\w\-]+)\s*$/){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
68 $chr = $1;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
69 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
70 else {
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
71 print STDERR "Unable to detect chromosome in pileup file\n$line\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
72 exit(0);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
73 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
74 if ($fields[1]=~/^\s*(\d+)\s*$/){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
75 $pos = $1;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
76 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
77 else {
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
78 print STDERR "Unable to detect position in pileup file\n$line\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
79 exit(0);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
80 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
81
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
82 if ($fields[2]=~/^\s*([ATGCNX])\s*$/i){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
83 $ref = $1;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
84 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
85 else {
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
86 print STDERR "Unable to detect reference base in pileup file\n$line\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
87 exit(0);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
88 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
89
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
90 if ($fields[3]=~/^\s*(\d+)\s*$/i){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
91 $depth = $1;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
92 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
93 else {
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
94 print STDERR "Unable to detect depth in pileup file\n$line\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
95 exit(0);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
96 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
97
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
98 $pile = $fields[4];
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
99 $pile =~ s/\$//g; #the read start at this position
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
100 $pile =~ s/\^.//g; #the read end at this position followed by quality char
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
101
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
102 if ($mpileupFiltred{"$chr#$pos"}) {
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
103 $pile = $mpileupFiltred{"$chr#$pos"};
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
104 $pile =~ s/\$//g; #the read start at this position
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
105 $pile =~ s/\^.//g; #the read end at this position followed by quality char
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
106 my $ori_pile = $pile;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
107
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
108 my @detail = split(//,$pile);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
109 my $current_in="";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
110 my $current_del="";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
111 my $current_length=0;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
112 my $noindel_pile="";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
113 my @IN;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
114 my @DEL;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
115
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
116 $noindel_pile = $pile;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
117 while ($noindel_pile =~ /\+(\d+)/){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
118 my $length = $1;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
119 $noindel_pile =~ s/(\+\d+[ATGCNX]{$length})//i ;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
120 push(@IN,$1);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
121 #print "test : $pile / $noindel_pile\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
122 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
123 while ($noindel_pile =~ /\-(\d+)/){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
124 my $length = $1;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
125 $noindel_pile =~ s/(\-\d+[ATGCNX]{$length})//i ;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
126 push(@DEL,$1);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
127 #print "test : $pile / $noindel_pile\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
128 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
129
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
130 my %variant_type;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
131
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
132 my $indel_pile = $ori_pile;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
133 $variant_type{"IN"} = ($indel_pile =~ s/\+//gi);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
134 $variant_type{"DEL"} = ($indel_pile =~ s/\-//gi);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
135
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
136 my $base_pile = $noindel_pile;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
137 $variant_type{"A"} = ($base_pile =~ s/a//gi);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
138 $variant_type{"T"} = ($base_pile =~ s/t//gi);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
139 $variant_type{"C"} = ($base_pile =~ s/c//gi);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
140 $variant_type{"G"} = ($base_pile =~ s/g//gi);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
141 $variant_type{"N"} = ($base_pile =~ s/n//gi);
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
142
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
143 my $top_number=0;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
144 my $top_variant="";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
145 foreach my $key (sort{$variant_type{$b}<=>$variant_type{$a}} keys %variant_type){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
146 if ($variant_type{$key} > 0){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
147 if ($variant_type{$key} > $top_number){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
148 $top_number = $variant_type{$key};
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
149 $top_variant = $key;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
150 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
151 elsif ($variant_type{$key} == $top_number){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
152 $top_variant .= "/$key";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
153 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
154 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
155
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
156 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
157 print "$chr\t$pos\t$ref\t$top_variant\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
158
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
159 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
160 elsif (($pile=~/[\+\-]/)||($pile=~/[ATGCN]/i)){
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
161 print "$chr\t$pos\t$ref\tX\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
162 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
163 else {
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
164 print "$chr\t$pos\t$ref\t$ref\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
165 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
166 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
167 else {
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
168 print STDERR "Unrecognized pileup format, need at least 5 columns\n$line\n";
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
169 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
170 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
171 }
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
172 close PU;
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
173
e94977a3dcad Uploaded
jnavarro
parents:
diff changeset
174