0
|
1 #!/usr/bin/perl -w
|
|
2 #Filename:
|
|
3 #Author: Tian Dongmei
|
|
4 #Email: tiandm@big.ac.cn
|
|
5 #Date: 2013/7/19
|
|
6 #Modified:
|
|
7 #Description:
|
|
8 my $version=1.00;
|
|
9
|
|
10 use strict;
|
|
11 use Getopt::Long;
|
|
12 use File::Basename;
|
|
13
|
|
14 my %opts;
|
|
15 GetOptions(\%opts,"i=s","ref=s","index:s","v:i","p:i","o=s","h");
|
|
16 if (!(defined $opts{i} and defined $opts{o} ) || defined $opts{h}) { #necessary arguments
|
|
17 &usage;
|
|
18 }
|
|
19
|
|
20 my $filein=$opts{'i'};
|
|
21 my $fileout=$opts{'o'};
|
|
22 unless ($fileout=~/\/$/) {$fileout.="/";}
|
|
23 my $rfam=$opts{'ref'};
|
|
24 my $mis=defined $opts{'v'}? $opts{'v'} : 0;
|
|
25 my $index=defined $opts{'index'} ? $opts{'index'} : "";
|
|
26 my $threads=defined $opts{'p'} ? $opts{'p'} : 1;
|
|
27
|
|
28
|
|
29 #my $time=time();
|
|
30
|
|
31 #my $mapdir=$fileout."/rfam_match_".$time;
|
|
32 my $mapdir=$fileout."/rfam_match";
|
|
33 mkdir $mapdir;
|
|
34 chdir $mapdir;
|
|
35 ###check genome index
|
|
36 if (-s $index.".1.ebwt") {
|
|
37 }else{
|
|
38 &checkACGT($rfam);
|
|
39 `bowtie-build $rfam rfam`;
|
|
40 $index="rfam";
|
|
41 }
|
|
42
|
|
43 #chdir "rfam_match_1397022331";
|
|
44 ### genome mapping
|
|
45 `bowtie -v $mis -f -p $threads -k 1 $index $filein --al rfam_mapped.fa --un rfam_not_mapped.fa > rfam_mapped.bwt`;
|
|
46
|
|
47 sub checkACGT{
|
|
48 my $string;
|
|
49 open IN,"<$rfam";
|
|
50 while (my $aline=<IN>) {
|
|
51 if ($aline!~/^>/) {
|
|
52 $aline=~s/U/T/gi;
|
|
53 }
|
|
54 $string .=$aline;
|
|
55 }
|
|
56 close IN;
|
|
57 $rfam=basename($rfam);
|
|
58 open OUT, ">$rfam";
|
|
59 print OUT $string;
|
|
60 close OUT;
|
|
61 }
|
|
62
|
|
63 sub usage{
|
|
64 print <<"USAGE";
|
|
65 Version $version
|
|
66 Usage:
|
|
67 $0 -i -o
|
|
68 options:
|
|
69 -i input file# input reads fasta/fastq file
|
|
70 -ref input file# rfam file, which do not contain miRNAs
|
|
71 -index file-prefix #(must be indexed by bowtie-build) The parameter
|
|
72 string must be the prefix of the bowtie index. For instance, if
|
|
73 the first indexed file is called 'h_sapiens_37_asm.1.ebwt' then
|
|
74 the prefix is 'h_sapiens_37_asm'.##can be null
|
|
75 -v <int> report end-to-end hits w/ <=v mismatches; ignore qualities,default 0;
|
|
76
|
|
77 -p/--threads <int> number of alignment threads to launch (default: 1)
|
|
78
|
|
79 -o output directory
|
|
80
|
|
81 -h help
|
|
82 USAGE
|
|
83 exit(1);
|
|
84 }
|
|
85
|