Mercurial > repos > ucsb-phylogenetics > osiris_phylogenetics
annotate phylogenies/genetree_read_placement.pl @ 0:5b9a38ec4a39 draft default tip
First commit of old repositories
author | osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu> |
---|---|
date | Tue, 11 Mar 2014 12:19:13 -0700 |
parents | |
children |
rev | line source |
---|---|
0
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
1 #! /usr/bin/perl -w |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
2 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
3 use strict; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
4 use warnings; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
5 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
6 ##For debugging command line pass, uncomment next |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
7 #for (my $i=0; $i < @ARGV; $i++){ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
8 # print "Parameter #$i ".$ARGV[$i]."\n\n"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
9 #} |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
10 #exit; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
11 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
12 my $newgenes=shift(@ARGV); #0 new genes to align |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
13 my $align = shift(@ARGV); #1 alignment program to use |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
14 my $path = shift(@ARGV); #2 path to tree and gene data |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
15 my $name = shift(@ARGV); #3 name of gene family |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
16 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
17 #If $newgenes has not hits, do not do read placement, just write tree with no hits |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
18 my $buffer; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
19 my $lines = 0; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
20 open(FILE, $newgenes) or die "Can't open `$newgenes': $!"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
21 while (sysread FILE, $buffer, 4096) { |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
22 $lines += ($buffer =~ tr/\n//); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
23 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
24 close FILE; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
25 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
26 if($lines < 1){ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
27 print "No hits found. Skipping read placement\n Tree copied to output.\n"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
28 system "cp $path.tre RAxML_labelledTree.EPA_TEST"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
29 system "cp $path.tre RAxML_originalLabelledTree.EPA_TEST"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
30 }else{ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
31 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
32 #First concatenate fasta files and align |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
33 system "cat $newgenes $path.fas > toalign.fas"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
34 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
35 if($align eq "MUSCLE"){ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
36 system "muscle -in toalign.fas -out aligned.fas"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
37 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
38 elsif($align eq "MAFFT") { |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
39 system "mafft --auto toalign.fas > aligned.fas"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
40 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
41 elsif($align eq "PRANK") { |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
42 system "prank -d=toalign.fas -o=aligned -f=fasta -F"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
43 system "mv aligned.2.fas aligned.fas"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
44 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
45 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
46 #convert to phylip format, uses seqConverter.pl |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
47 system "/home/galaxy/galaxy-dist/tools/oakley_dev/seqConverterG.pl -daligned.fas -ope -Oaligned.phy"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
48 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
49 system "raxmlHPC-PTHREADS-SSE3 -f v -s aligned.phy -m PROTGAMMAWAG -t $path.tre -n EPA_TEST -T 4"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
50 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
51 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
52 #Now make tab delimited file to use in tab2trees |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
53 #open treefile to read tree line |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
54 open(TREE, "<","RAxML_labelledTree.EPA_TEST") or die "Can't open RESULT File!"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
55 my $finaltree; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
56 while (<TREE>){ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
57 if($_ =~ /\;/m){ |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
58 $finaltree = $_; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
59 chomp($finaltree); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
60 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
61 } |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
62 close TREE; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
63 |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
64 $name =~ s/ /_/g; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
65 chomp($name); |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
66 #remove clade labels |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
67 $finaltree =~ s/\[I\d+\]//g; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
68 open(TAB, '>treeout.tab') or die "Can't open File!"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
69 print TAB $name."\t".$finaltree."\n"; |
5b9a38ec4a39
First commit of old repositories
osiris_phylogenetics <ucsb_phylogenetics@lifesci.ucsb.edu>
parents:
diff
changeset
|
70 close TAB; |