Mercurial > repos > mir-bioinf > multi_join_left

annotate multi_join_left/addColumnsFromFile2ToFile1.pl @ 5:1de2a8f041b3 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
author mir-bioinf
date Tue, 21 Apr 2015 16:20:03 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
1 #! /usr/bin/perl -w
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
2 #===============================================================================
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
3 #
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
4 # FILENAME: addColumnsFromFile2ToFile1.pl
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
5 #
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
6 # USAGE: see -help
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
7 #
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
8 # DESCRIPTION: This program adds columns in File 2 to File 1,
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
9 # if there are correnponding entries in File 1
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
10 #
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
11 # AUTHOR: Ron Stewart
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
12 # VERSION: 1.1
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
13 # CREATED: 12/18/06 CDT
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
14 #===============================================================================
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
15
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
16 use lib '/opt/galaxy/galaxy-dist/tools/ngs_rna/Unreleased'; ##CMS ADDED 11-05-14, DIR CONTAINS CUSTOM MODULE
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
17
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
18 use strict;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
19 use IO::File;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
20 use GetOptWC;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
21 # parsing the command line
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
22 my %optVarsIn = ();
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
23 # help information
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
24 $optVarsIn{'File1'} = './File1.txt`=s`Input file of genes to include';
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
25 $optVarsIn{'File2'} = './File2.txt`=s`Input file to be added';
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
26 $optVarsIn{'cola1'} = '2`=f`first column to look for match in File1';
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
27 $optVarsIn{'colb1'} = '4`=f`second column to look for match in File1';
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
28 $optVarsIn{'cola2'} = '2`=f`first column to look for match in File2';
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
29 $optVarsIn{'colb2'} = '4`=f`second column to look for match in File2';
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
30 $optVarsIn{'file1FirstColToCopy'} = '-1`=f`first column to copy in File1';
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
31 $optVarsIn{'file1LastColToCopy'} = '-1`=f`last column to copy in File1';
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
32 $optVarsIn{'file2FirstColToCopy'} = '-1`=f`first column to copy in File2';
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
33 $optVarsIn{'file2LastColToCopy'} = '-1`=f`last column to copy in File2';
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
34 $optVarsIn{'HelpPrefix'} = 'This script is for adding entries in File2 to the corresponding entries in File1.';
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
35 $optVarsIn{'HelpSuffix'} = 'example call: ./addColumnsFromFile2ToFile1.pl -File1=./esAndDiffMarkersWithSage20061211.txt -cola1=2 -calb1=4 -File2=./pan_whole_table_fold_ann.txt -cola2=1 -colb2=4`Note: Two input files should have title lines`NOTE:Files must be text files, NOT .xls files. If you have an .xls file, save it as "Text(Windows) in Excel.';
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
36 my %retVars = ();
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
37 my $retVarsRef = GetOptWC::getOptions(\%optVarsIn);
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
38 %retVars = %$retVarsRef;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
39 if ($retVars{'HelpCalled'}) {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
40 print "exiting now, help called\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
41 exit;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
42 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
43 my $File1 = $retVars{'File1'};
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
44 my $FHFile1;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
45 $FHFile1 = IO::File->new("<$File1");
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
46
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
47 my $File2 = $retVars{'File2'};
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
48 my $FHFile2;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
49 $FHFile2 = IO::File->new("<$File2");
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
50 my $File1name = $File1;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
51 print "file1name: $File1name\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
52 $File1name =~ s/[\.\/]/_/g;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
53 print "file1name: $File1name\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
54
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
55 my $File2name = $File2;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
56 $File2name =~ s/[\.\/]/_/g;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
57 #my $Out = $File1name.'.'.$File2name; # this can be too long in some cases
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
58 my $Out = "file1_file2.txt";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
59 print" out is $Out\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
60 my $OutFile = IO::File->new(">$Out");
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
61 my $cola1 = $retVars{'cola1'};
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
62 my $colb1 = $retVars{'colb1'};
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
63 my $cola2 = $retVars{'cola2'};
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
64 my $colb2 = $retVars{'colb2'};
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
65 my $firstColFile1 = $retVars{'file1FirstColToCopy'};
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
66 my $lastColFile1 = $retVars{'file1LastColToCopy'};
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
67 my $firstColFile2 = $retVars{'file2FirstColToCopy'};
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
68 my $lastColFile2 = $retVars{'file2LastColToCopy'};
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
69 my %genes2 = ();
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
70 my %genes4 = ();
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
71 my %genes4key = ();
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
72 my $lineCtr = 0;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
73 my @cols = ();
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
74 my $firstLineFile2 = $FHFile2->getline();
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
75 $firstLineFile2 =~ s/\s+$//;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
76 @cols = split "\t",$firstLineFile2;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
77 my $numColFile2 = @cols;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
78 if($firstColFile2==-1){
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
79 $firstColFile2 = 0;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
80 $lastColFile2 = $numColFile2-1;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
81 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
82 my @titleFile2 = @cols[$firstColFile2..$lastColFile2];
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
83 while (my $line = $FHFile2->getline()) {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
84 $lineCtr++;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
85 #$line =~ s/\s+$//;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
86 $line =~ s/\R//g; ##CMS 11-6-14
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
87 chomp($line);
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
88 @cols = split "\t",$line;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
89 my $numCols = (@cols + 0);
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
90
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
91 #$cols[$cola2] = uc($cols[$cola2]); ##CMS 11-6-14
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
92 #$cols[$colb2] = uc($cols[$colb2]); ##CMS 11-6-14
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
93 $cols[$cola2] =~ s/ //g;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
94 my $colsBSymbol = "";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
95 if ($cols[$colb2] =~ /\"{0,1}CDS\; ([^\;]+);/) {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
96 $colsBSymbol = $1;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
97 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
98 else {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
99 #if ($numCols == ($colb2 +1)) { ##CMS COMMENTED OUT 11-6-14
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
100 #print "$line\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
101 #print "Please check this line\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
102 #exit;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
103 #} ##CMS END COMMENTS 11-6-14
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
104 $cols[$colb2] =~ s/ //g;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
105 $colsBSymbol = $cols[$colb2];
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
106 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
107 if($numCols<$lastColFile2){
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
108 for(my $i = $numCols;$i<$lastColFile2;$i++){
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
109 $cols[$i] = "";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
110 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
111 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
112
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
113 $genes2{$cols[$cola2]} = join("\t",@cols[$firstColFile2..$lastColFile2]);
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
114 $genes4{$colsBSymbol}->{$cols[$cola2]} = join("\t",@cols[$firstColFile2..$lastColFile2]);
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
115 $genes4key{$colsBSymbol}="x";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
116 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
117 print "linectr: $lineCtr\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
118 $lineCtr = 0;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
119 @cols = ();
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
120 my $firstLineFile1 = $FHFile1->getline();
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
121 $firstLineFile1 =~ s/\s+$//;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
122 @cols = split "\t",$firstLineFile1;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
123 my $numColFile1 = @cols;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
124 if($firstColFile1==-1){
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
125 $firstColFile1 = 0;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
126 $lastColFile1 = $numColFile1-1;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
127 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
128 #print "numcolsfile1: $numColFile1\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
129 #print "lastcolsfile1: $lastColFile1\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
130
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
131 my @titleFile1 = @cols[$firstColFile1..$lastColFile1];
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
132 #print "tf1: @titleFile1\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
133 #print "tf2: @titleFile2\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
134 #print "outfile: $OutFile\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
135
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
136 print $OutFile join("\t",@titleFile1)."\t".join("\t",@titleFile2)."\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
137 #my $numCol = $lastColFile1 - $firstColFile1 +1;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
138 my $numCurrentLine =0;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
139 while (my $line = $FHFile1->getline()) {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
140 $lineCtr++;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
141 $line =~ s/\s+$//;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
142 my $selectedEntries;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
143 @cols = split "\t",$line;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
144 $numCurrentLine = $#cols;#[$firstColFile1..$LastColFile1];
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
145 #print "numcurrentline: $numCurrentLine\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
146 $line = $line."\t";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
147 if($numCurrentLine<$lastColFile1){
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
148 #print "in if\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
149 for(my $i =$numCurrentLine+1;$i<=$lastColFile1;$i++){
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
150 #print "in for. i=$i\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
151 $cols[$i]="";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
152 #$line = $line."\t";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
153 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
154 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
155 $selectedEntries = join("\t",@cols[$firstColFile1..$lastColFile1]);
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
156 my $numCols = (@cols + 0);
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
157 #$cols[$cola1] = uc($cols[$cola1]); ##CMS 11-6-14
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
158 #$cols[$colb1] = uc($cols[$colb1]); ##CMS 11-6-14
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
159 $cols[$cola1] =~ s/ //g;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
160 my $colsBSymbol = "";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
161 if ($cols[$colb1] =~ /\"{0,1}CDS\; ([^\;]+);/) {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
162 $colsBSymbol = $1;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
163 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
164 else {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
165 #if ($numCols == ($colb1 +1)) { ##CMS COMMENTED 11-6-14
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
166 # print"$line";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
167 # print " please check this line\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
168 # exit;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
169 #} ##CMS END COMMENTS 11-6-14
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
170 $cols[$colb1] =~ s/ //g;
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
171 $colsBSymbol = $cols[$colb1];
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
172 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
173 if((exists ($genes2{$cols[$cola1]}) and $cols[$cola1] ne "N/A") or (exists ($genes2{$colsBSymbol}) and $colsBSymbol ne "N/A") ) {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
174 if (exists ($genes2{$cols[$cola1]})) {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
175 print $OutFile "$selectedEntries"."\t".$genes2{$cols[$cola1]}."\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
176 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
177 elsif (exists ($genes2{$colsBSymbol})) {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
178 print $OutFile "$selectedEntries"."\t".$genes2{$colsBSymbol}."\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
179 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
180 else {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
181 print "WHOA, we've got a problem Here!!!!!\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
182 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
183 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
184 elsif(exists ($genes4key{$colsBSymbol}) and $colsBSymbol ne "N/A" ) {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
185 foreach my $symbol (keys %{$genes4{$colsBSymbol}}){
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
186 print $OutFile "$selectedEntries"."\t".$genes4{$colsBSymbol}->{$symbol}."\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
187 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
188 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
189 else {
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
190 print $OutFile "$selectedEntries"."\n";
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
191 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
192 }
1de2a8f041b3 Corrected xml syntax errors in config file, added missing pl script and output dataset to test-data
mir-bioinf
parents:
diff changeset
193 exit;