annotate variant_effect_predictor/Bio/EnsEMBL/Funcgen/RunnableDB/InferMotifs.pm @ 2:a5976b2dce6f

changing defualt values for ensembl database
author mahtabm
date Thu, 11 Apr 2013 17:15:42 +1000
parents 1f6dce3d34e0
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
1 =pod
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
2
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
3 =head1 NAME
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
4
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
5 Bio::EnsEMBL::Funcgen::RunnableDB::InferMotifs
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
6
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
7 =head1 DESCRIPTION
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
8
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
9 'Funcgen::InferMotifs'
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
10
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
11 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
12
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
13
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
14 package Bio::EnsEMBL::Funcgen::RunnableDB::InferMotifs;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
15
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
16 use warnings;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
17 use strict;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
18
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
19 use base ('Bio::EnsEMBL::Funcgen::RunnableDB::Motif');
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
20
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
21 use Bio::EnsEMBL::Utils::Exception qw(throw warning stack_trace_dump);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
22 use Bio::EnsEMBL::Funcgen::Utils::EFGUtils qw (run_system_cmd);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
23 use Data::Dumper;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
24
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
25 sub fetch_input { # nothing to fetch... just the parameters...
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
26 my $self = shift @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
27
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
28 $self->SUPER::fetch_input();
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
29
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
30 if(!$self->param('bin')){ throw "No bin defined!"; }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
31 $self->_bin($self->param('bin'));
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
32
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
33 return 1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
34 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
35
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
36 sub run {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
37
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
38 my $self = shift @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
39
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
40 my $afa = $self->_efgdba()->get_AnnotatedFeatureAdaptor();
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
41 #Order features descending by score
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
42 my @features = sort { $b->score <=> $a->score } @{$afa->fetch_all_by_FeatureSets( [ $self->_feature_set ] )};
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
43
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
44 # Print the sequences for this bin
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
45 my $sa = $self->_efgdba()->dnadb->get_SliceAdaptor();
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
46 my $fasta_file = $self->_output_dir."/bin.".$self->_bin.".fasta";
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
47 open(FO,">".$fasta_file) or throw "Couldn't open output file";
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
48 my $bin_start = ($self->_bin - 1)*$self->_bin_size;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
49 for(my $i=$bin_start; $i<($bin_start+$self->_bin_size); $i++){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
50 my $ft = $features[$i];
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
51 my $sr_name = $ft->seq_region_name;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
52 my $start = POSIX::floor($ft->summit - $self->_window_size);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
53 my $end = POSIX::floor($ft->summit + $self->_window_size);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
54 my $slice = $sa->fetch_by_region( undef, $sr_name, $start, $end);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
55 if(defined($slice)){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
56 print FO ">".$sr_name.",".$start.",".$end."\n".$slice->get_repeatmasked_seq()->seq."\n";
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
57 } else {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
58 warn $sr_name.",".$start.",".$end." could not be found\n";
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
59 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
60 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
61 close FO;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
62
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
63 #Create a background to try and avoid repetitive A's
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
64 #fasta-get-markov is part of the meme package (but is very simple to replicate if needed)
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
65 #Use a pre-compiled background unique for the whole of human...
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
66 #run_system_cmd("cat ".$fasta_file." | fasta-get-markov -m 3 > ".$fasta_file.".bg");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
67
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
68 #Run MEME for this bin
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
69 my $species = $self->_species;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
70 my $meme_file = $self->_output_dir."/bin.".$self->_bin.".meme";
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
71 #Replace meme parameters by parameters from an analysis in the database...
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
72 #run_system_cmd("meme.bin -nostatus -dna -text -revcomp -mod zoops -evt 0.00001 -nmotifs 10 -minsites 50 -minw 6 -maxw 20 -bfile ~/src/STAMP.v1.1/tests/${species}_masked.bg $fasta_file > ".$meme_file);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
73 run_system_cmd($self->_bin_folder()."/meme.bin -nostatus -dna -text -revcomp -mod zoops -evt 1e-20 -nmotifs 5 -minsites 50 -minw 6 -maxw 20 $fasta_file > ".$meme_file);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
74
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
75 my $basename = $self->_feature_set->name.".bin_".$self->_bin;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
76 if(_process_meme_to_STAMP($meme_file, $basename) == 0){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
77 warn "No motif found ";
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
78 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
79
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
80 #Eliminate the temp files...
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
81 #run_system_cmd("rm -f ".$fasta_file);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
82 #run_system_cmd("rm -f ".$meme_file);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
83
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
84 return 1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
85 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
86
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
87
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
88 sub write_output { # Nothing is written at this stage (for the moment)
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
89
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
90 my $self = shift @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
91
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
92 return 1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
93
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
94 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
95
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
96
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
97 sub _process_meme_to_STAMP {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
98 my ($meme_file, $basename) = (shift, shift, shift);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
99
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
100 my $num_motifs = 0;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
101 my $log_odds = 1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
102 open(FO,">".$meme_file.".tmp_TRANSFAC");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
103 open(FILE, $meme_file);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
104 while(<FILE>){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
105 chomp;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
106 if(/^MOTIF\s+.*E-value\s*=\s+(\S+)/) { $log_odds = $1; }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
107 if(/^BL\s+MOTIF\s+(\d+)\s+width=(\d+)\s+seqs=(\d+)\s*$/){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
108 my $motif_num = $1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
109 my $motif_width = $2;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
110 my $num_seqs = $3;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
111
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
112
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
113 my @init = (0) x $motif_width;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
114 my %matrix;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
115 #Initialize with 0
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
116 for (my $j=0; $j<$motif_width; $j++){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
117 $matrix{"A"}->[$j] = $matrix{"C"}->[$j] = $matrix{"G"}->[$j] = $matrix{"T"}->[$j] = 0;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
118 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
119
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
120 for (my $i=1;$i<=$num_seqs;$i++){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
121 my $line = <FILE>;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
122 $line =~ /^\S+\s+\(\s*\d+\)\s+(\w+)\s+\d+\s*$/;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
123 my @seq = split(//,uc($1));
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
124 if(scalar(@seq) != $motif_width){ throw "Problems parsing motif file"; }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
125 for (my $j=0; $j<$motif_width; $j++){ $matrix{$seq[$j]}->[$j]++ }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
126 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
127 my $end = <FILE>;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
128 if(! $end =~ /^\/\/\s*$/){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
129 throw "Error in the MEME file: expecting \/\/ after $num_seqs lines ";
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
130 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
131 print FO "DE ".$basename.".".$motif_num."\t".$log_odds."\n";
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
132 for(my $i=0; $i<$motif_width; $i++){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
133 print FO $i."\t".$matrix{"A"}->[$i]."\t".$matrix{"C"}->[$i]."\t".$matrix{"G"}->[$i]."\t".$matrix{"T"}->[$i]."\n"
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
134 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
135 print FO "XX\n";
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
136
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
137 $num_motifs++;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
138
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
139 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
140
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
141 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
142 close FILE;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
143 close FO;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
144
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
145 return $num_motifs;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
146
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
147 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
148
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
149 #Private getter / setter to the bin
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
150 sub _bin {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
151 return $_[0]->_getter_setter('bin',$_[1]);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
152 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
153
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
154 1;