Mercurial > repos > mir-bioinf > multi_join_left

comparison 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
comparison
equal deleted inserted replaced
4:46c880ae6db2 5:1de2a8f041b3
1 #! /usr/bin/perl -w
2 #===============================================================================
3 #
4 # FILENAME: addColumnsFromFile2ToFile1.pl
5 #
6 # USAGE: see -help
7 #
8 # DESCRIPTION: This program adds columns in File 2 to File 1,
9 # if there are correnponding entries in File 1
10 #
11 # AUTHOR: Ron Stewart
12 # VERSION: 1.1
13 # CREATED: 12/18/06 CDT
14 #===============================================================================
15
16 use lib '/opt/galaxy/galaxy-dist/tools/ngs_rna/Unreleased'; ##CMS ADDED 11-05-14, DIR CONTAINS CUSTOM MODULE
17
18 use strict;
19 use IO::File;
20 use GetOptWC;
21 # parsing the command line
22 my %optVarsIn = ();
23 # help information
24 $optVarsIn{'File1'} = './File1.txt`=s`Input file of genes to include';
25 $optVarsIn{'File2'} = './File2.txt`=s`Input file to be added';
26 $optVarsIn{'cola1'} = '2`=f`first column to look for match in File1';
27 $optVarsIn{'colb1'} = '4`=f`second column to look for match in File1';
28 $optVarsIn{'cola2'} = '2`=f`first column to look for match in File2';
29 $optVarsIn{'colb2'} = '4`=f`second column to look for match in File2';
30 $optVarsIn{'file1FirstColToCopy'} = '-1`=f`first column to copy in File1';
31 $optVarsIn{'file1LastColToCopy'} = '-1`=f`last column to copy in File1';
32 $optVarsIn{'file2FirstColToCopy'} = '-1`=f`first column to copy in File2';
33 $optVarsIn{'file2LastColToCopy'} = '-1`=f`last column to copy in File2';
34 $optVarsIn{'HelpPrefix'} = 'This script is for adding entries in File2 to the corresponding entries in File1.';
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.';
36 my %retVars = ();
37 my $retVarsRef = GetOptWC::getOptions(\%optVarsIn);
38 %retVars = %$retVarsRef;
39 if ($retVars{'HelpCalled'}) {
40 print "exiting now, help called\n";
41 exit;
42 }
43 my $File1 = $retVars{'File1'};
44 my $FHFile1;
45 $FHFile1 = IO::File->new("<$File1");
46
47 my $File2 = $retVars{'File2'};
48 my $FHFile2;
49 $FHFile2 = IO::File->new("<$File2");
50 my $File1name = $File1;
51 print "file1name: $File1name\n";
52 $File1name =~ s/[\.\/]/_/g;
53 print "file1name: $File1name\n";
54
55 my $File2name = $File2;
56 $File2name =~ s/[\.\/]/_/g;
57 #my $Out = $File1name.'.'.$File2name; # this can be too long in some cases
58 my $Out = "file1_file2.txt";
59 print" out is $Out\n";
60 my $OutFile = IO::File->new(">$Out");
61 my $cola1 = $retVars{'cola1'};
62 my $colb1 = $retVars{'colb1'};
63 my $cola2 = $retVars{'cola2'};
64 my $colb2 = $retVars{'colb2'};
65 my $firstColFile1 = $retVars{'file1FirstColToCopy'};
66 my $lastColFile1 = $retVars{'file1LastColToCopy'};
67 my $firstColFile2 = $retVars{'file2FirstColToCopy'};
68 my $lastColFile2 = $retVars{'file2LastColToCopy'};
69 my %genes2 = ();
70 my %genes4 = ();
71 my %genes4key = ();
72 my $lineCtr = 0;
73 my @cols = ();
74 my $firstLineFile2 = $FHFile2->getline();
75 $firstLineFile2 =~ s/\s+$//;
76 @cols = split "\t",$firstLineFile2;
77 my $numColFile2 = @cols;
78 if($firstColFile2==-1){
79 $firstColFile2 = 0;
80 $lastColFile2 = $numColFile2-1;
81 }
82 my @titleFile2 = @cols[$firstColFile2..$lastColFile2];
83 while (my $line = $FHFile2->getline()) {
84 $lineCtr++;
85 #$line =~ s/\s+$//;
86 $line =~ s/\R//g; ##CMS 11-6-14
87 chomp($line);
88 @cols = split "\t",$line;
89 my $numCols = (@cols + 0);
90
91 #$cols[$cola2] = uc($cols[$cola2]); ##CMS 11-6-14
92 #$cols[$colb2] = uc($cols[$colb2]); ##CMS 11-6-14
93 $cols[$cola2] =~ s/ //g;
94 my $colsBSymbol = "";
95 if ($cols[$colb2] =~ /\"{0,1}CDS\; ([^\;]+);/) {
96 $colsBSymbol = $1;
97 }
98 else {
99 #if ($numCols == ($colb2 +1)) { ##CMS COMMENTED OUT 11-6-14
100 #print "$line\n";
101 #print "Please check this line\n";
102 #exit;
103 #} ##CMS END COMMENTS 11-6-14
104 $cols[$colb2] =~ s/ //g;
105 $colsBSymbol = $cols[$colb2];
106 }
107 if($numCols<$lastColFile2){
108 for(my $i = $numCols;$i<$lastColFile2;$i++){
109 $cols[$i] = "";
110 }
111 }
112
113 $genes2{$cols[$cola2]} = join("\t",@cols[$firstColFile2..$lastColFile2]);
114 $genes4{$colsBSymbol}->{$cols[$cola2]} = join("\t",@cols[$firstColFile2..$lastColFile2]);
115 $genes4key{$colsBSymbol}="x";
116 }
117 print "linectr: $lineCtr\n";
118 $lineCtr = 0;
119 @cols = ();
120 my $firstLineFile1 = $FHFile1->getline();
121 $firstLineFile1 =~ s/\s+$//;
122 @cols = split "\t",$firstLineFile1;
123 my $numColFile1 = @cols;
124 if($firstColFile1==-1){
125 $firstColFile1 = 0;
126 $lastColFile1 = $numColFile1-1;
127 }
128 #print "numcolsfile1: $numColFile1\n";
129 #print "lastcolsfile1: $lastColFile1\n";
130
131 my @titleFile1 = @cols[$firstColFile1..$lastColFile1];
132 #print "tf1: @titleFile1\n";
133 #print "tf2: @titleFile2\n";
134 #print "outfile: $OutFile\n";
135
136 print $OutFile join("\t",@titleFile1)."\t".join("\t",@titleFile2)."\n";
137 #my $numCol = $lastColFile1 - $firstColFile1 +1;
138 my $numCurrentLine =0;
139 while (my $line = $FHFile1->getline()) {
140 $lineCtr++;
141 $line =~ s/\s+$//;
142 my $selectedEntries;
143 @cols = split "\t",$line;
144 $numCurrentLine = $#cols;#[$firstColFile1..$LastColFile1];
145 #print "numcurrentline: $numCurrentLine\n";
146 $line = $line."\t";
147 if($numCurrentLine<$lastColFile1){
148 #print "in if\n";
149 for(my $i =$numCurrentLine+1;$i<=$lastColFile1;$i++){
150 #print "in for. i=$i\n";
151 $cols[$i]="";
152 #$line = $line."\t";
153 }
154 }
155 $selectedEntries = join("\t",@cols[$firstColFile1..$lastColFile1]);
156 my $numCols = (@cols + 0);
157 #$cols[$cola1] = uc($cols[$cola1]); ##CMS 11-6-14
158 #$cols[$colb1] = uc($cols[$colb1]); ##CMS 11-6-14
159 $cols[$cola1] =~ s/ //g;
160 my $colsBSymbol = "";
161 if ($cols[$colb1] =~ /\"{0,1}CDS\; ([^\;]+);/) {
162 $colsBSymbol = $1;
163 }
164 else {
165 #if ($numCols == ($colb1 +1)) { ##CMS COMMENTED 11-6-14
166 # print"$line";
167 # print " please check this line\n";
168 # exit;
169 #} ##CMS END COMMENTS 11-6-14
170 $cols[$colb1] =~ s/ //g;
171 $colsBSymbol = $cols[$colb1];
172 }
173 if((exists ($genes2{$cols[$cola1]}) and $cols[$cola1] ne "N/A") or (exists ($genes2{$colsBSymbol}) and $colsBSymbol ne "N/A") ) {
174 if (exists ($genes2{$cols[$cola1]})) {
175 print $OutFile "$selectedEntries"."\t".$genes2{$cols[$cola1]}."\n";
176 }
177 elsif (exists ($genes2{$colsBSymbol})) {
178 print $OutFile "$selectedEntries"."\t".$genes2{$colsBSymbol}."\n";
179 }
180 else {
181 print "WHOA, we've got a problem Here!!!!!\n";
182 }
183 }
184 elsif(exists ($genes4key{$colsBSymbol}) and $colsBSymbol ne "N/A" ) {
185 foreach my $symbol (keys %{$genes4{$colsBSymbol}}){
186 print $OutFile "$selectedEntries"."\t".$genes4{$colsBSymbol}->{$symbol}."\n";
187 }
188 }
189 else {
190 print $OutFile "$selectedEntries"."\n";
191 }
192 }
193 exit;