annotate snpEff_2_1a/snpEff_2_1a/scripts/fastaSample.pl @ 3:c052639fa666 default tip

Uploaded
author pcingola
date Fri, 20 Apr 2012 11:22:59 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3
c052639fa666 Uploaded
pcingola
parents:
diff changeset
1 #!/usr/bin/perl
c052639fa666 Uploaded
pcingola
parents:
diff changeset
2
c052639fa666 Uploaded
pcingola
parents:
diff changeset
3 #---
c052639fa666 Uploaded
pcingola
parents:
diff changeset
4 # Initialize parameters
c052639fa666 Uploaded
pcingola
parents:
diff changeset
5 #---
c052639fa666 Uploaded
pcingola
parents:
diff changeset
6 $sampleStart = $ARGV[0];
c052639fa666 Uploaded
pcingola
parents:
diff changeset
7 $sampleEnd = $ARGV[1];
c052639fa666 Uploaded
pcingola
parents:
diff changeset
8 if(( $ARGV[0] eq '' ) || ($ARGV[1] eq '')) { die "Usage: fastaSample sampleStart sampleEnd\n"; }
c052639fa666 Uploaded
pcingola
parents:
diff changeset
9
c052639fa666 Uploaded
pcingola
parents:
diff changeset
10 $sampleStart--;
c052639fa666 Uploaded
pcingola
parents:
diff changeset
11 $sampleEnd--;
c052639fa666 Uploaded
pcingola
parents:
diff changeset
12 $sampleLen = $sampleEnd - $sampleStart + 1;
c052639fa666 Uploaded
pcingola
parents:
diff changeset
13
c052639fa666 Uploaded
pcingola
parents:
diff changeset
14 #---
c052639fa666 Uploaded
pcingola
parents:
diff changeset
15 # Read fasta file
c052639fa666 Uploaded
pcingola
parents:
diff changeset
16 #---
c052639fa666 Uploaded
pcingola
parents:
diff changeset
17 for($lineNum=0 ; $l = <STDIN> ; $lineNum++ ) {
c052639fa666 Uploaded
pcingola
parents:
diff changeset
18 if( $l =~/^>/ ) {
c052639fa666 Uploaded
pcingola
parents:
diff changeset
19 # Sample if not empty
c052639fa666 Uploaded
pcingola
parents:
diff changeset
20 if( $seq ne "" ) {
c052639fa666 Uploaded
pcingola
parents:
diff changeset
21 $s = substr( $seq, $sampleStart, $sampleLen);
c052639fa666 Uploaded
pcingola
parents:
diff changeset
22 print "$s\n";
c052639fa666 Uploaded
pcingola
parents:
diff changeset
23 }
c052639fa666 Uploaded
pcingola
parents:
diff changeset
24 # New sequence
c052639fa666 Uploaded
pcingola
parents:
diff changeset
25 $seq = "";
c052639fa666 Uploaded
pcingola
parents:
diff changeset
26 } else {
c052639fa666 Uploaded
pcingola
parents:
diff changeset
27 chomp($l);
c052639fa666 Uploaded
pcingola
parents:
diff changeset
28 $seq .= $l;
c052639fa666 Uploaded
pcingola
parents:
diff changeset
29 }
c052639fa666 Uploaded
pcingola
parents:
diff changeset
30 }
c052639fa666 Uploaded
pcingola
parents:
diff changeset
31
c052639fa666 Uploaded
pcingola
parents:
diff changeset
32 # Sample if not empty
c052639fa666 Uploaded
pcingola
parents:
diff changeset
33 if( $seq ne "" ) {
c052639fa666 Uploaded
pcingola
parents:
diff changeset
34 $s = substr( $seq, $sampleStart, $sampleLen);
c052639fa666 Uploaded
pcingola
parents:
diff changeset
35 print "$s\n";
c052639fa666 Uploaded
pcingola
parents:
diff changeset
36 }
c052639fa666 Uploaded
pcingola
parents:
diff changeset
37
c052639fa666 Uploaded
pcingola
parents:
diff changeset
38 $len = length($seq);
c052639fa666 Uploaded
pcingola
parents:
diff changeset
39 print STDERR "Lines: $lineNum\n";
c052639fa666 Uploaded
pcingola
parents:
diff changeset
40 print STDERR "Sequence size: $len\n";
c052639fa666 Uploaded
pcingola
parents:
diff changeset
41 print STDERR "Sample size: $sampleLen\n";