annotate rfam.pl @ 53:f5a2e8308836 draft default tip

Uploaded
author big-tiandm
date Mon, 08 Dec 2014 01:51:16 -0500
parents c75593f79aa9
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
37
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
1 #!/usr/bin/perl -w
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
2 #Filename:
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
3 #Author: Tian Dongmei
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
4 #Email: tiandm@big.ac.cn
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
5 #Date: 2013/7/19
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
6 #Modified:
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
7 #Description:
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
8 my $version=1.00;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
9
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
10 use strict;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
11 use Getopt::Long;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
12 use File::Basename;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
13
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
14 my %opts;
47
c75593f79aa9 Uploaded
big-tiandm
parents: 44
diff changeset
15 GetOptions(\%opts,"i=s","ref=s","index:s","v:i","p:i","o=s","h");
37
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
16 if (!(defined $opts{i} and defined $opts{o} ) || defined $opts{h}) { #necessary arguments
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
17 &usage;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
18 }
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
19
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
20 my $filein=$opts{'i'};
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
21 my $dir=$opts{'o'};
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
22 unless ($dir=~/\/$/) {$dir.="/";}
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
23 my $rfam=$opts{'ref'};
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
24 my $mis=defined $opts{'v'}? $opts{'v'} : 0;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
25 my $index=defined $opts{'index'} ? $opts{'index'} : "";
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
26 my $threads=defined $opts{'p'} ? $opts{'p'} : 1;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
27
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
28 if (not -d $dir) {
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
29 mkdir $dir;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
30 }
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
31
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
32
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
33 my $time=Time();
47
c75593f79aa9 Uploaded
big-tiandm
parents: 44
diff changeset
34
c75593f79aa9 Uploaded
big-tiandm
parents: 44
diff changeset
35 my $mapdir=$dir."/rfam_match";
37
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
36 if(not -d $mapdir){
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
37 mkdir $mapdir;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
38 }
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
39 chdir $mapdir;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
40 ###check genome index
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
41 if (-s $index.".1.ebwt") {
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
42 }else{
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
43 &checkACGT($rfam);
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
44 `bowtie-build $rfam $rfam`;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
45 $index="$rfam";
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
46 }
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
47 ### genome mapping
44
0c4e11018934 Uploaded
big-tiandm
parents: 37
diff changeset
48 `bowtie -v $mis -f -p $threads -k 1 $index $filein --al rfam_mapped.fa --un rfam_not_mapped.fa > rfam_mapped.bwt 2> run.log`;
37
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
49
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
50 sub checkACGT{
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
51 my $string;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
52 open IN,"<$rfam";
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
53 while (my $aline=<IN>) {
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
54 if ($aline!~/^>/) {
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
55 $aline=~s/U/T/gi;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
56 }
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
57 $string .=$aline;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
58 }
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
59 close IN;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
60 $rfam=basename($rfam);
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
61 open OUT, ">$rfam";
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
62 print OUT $string;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
63 close OUT;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
64 }
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
65
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
66 sub Time{
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
67 my $time=time();
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
68 my ($sec,$min,$hour,$day,$month,$year) = (localtime($time))[0,1,2,3,4,5,6];
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
69 $month++;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
70 $year+=1900;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
71 if (length($sec) == 1) {$sec = "0"."$sec";}
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
72 if (length($min) == 1) {$min = "0"."$min";}
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
73 if (length($hour) == 1) {$hour = "0"."$hour";}
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
74 if (length($day) == 1) {$day = "0"."$day";}
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
75 if (length($month) == 1) {$month = "0"."$month";}
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
76 #print "$year-$month-$day $hour:$min:$sec\n";
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
77 return("$year-$month-$day-$hour-$min-$sec");
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
78 }
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
79 sub usage{
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
80 print <<"USAGE";
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
81 Version $version
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
82 Usage:
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
83 $0 -i -o
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
84 options:
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
85 -i input file# input reads fasta/fastq file
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
86 -ref input file# rfam file, which do not contain miRNAs
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
87 -index file-prefix #(must be indexed by bowtie-build) The parameter
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
88 string must be the prefix of the bowtie index. For instance, if
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
89 the first indexed file is called 'h_sapiens_37_asm.1.ebwt' then
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
90 the prefix is 'h_sapiens_37_asm'.##can be null
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
91 -v <int> report end-to-end hits w/ <=v mismatches; ignore qualities,default 0;
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
92
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
93 -p/--threads <int> number of alignment threads to launch (default: 1)
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
94
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
95 -o output directory
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
96 -h help
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
97 USAGE
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
98 exit(1);
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
99 }
9ae0d25e4169 Uploaded
big-tiandm
parents:
diff changeset
100