|
0
|
1 #!/usr/bin/perl -w
|
|
|
2 #
|
|
|
3 # $RCSfile: ElementalAnalysisTextFiles.pl,v $
|
|
|
4 # $Date: 2015/02/28 20:46:19 $
|
|
|
5 # $Revision: 1.28 $
|
|
|
6 #
|
|
|
7 # Author: Manish Sud <msud@san.rr.com>
|
|
|
8 #
|
|
|
9 # Copyright (C) 2015 Manish Sud. All rights reserved.
|
|
|
10 #
|
|
|
11 # This file is part of MayaChemTools.
|
|
|
12 #
|
|
|
13 # MayaChemTools is free software; you can redistribute it and/or modify it under
|
|
|
14 # the terms of the GNU Lesser General Public License as published by the Free
|
|
|
15 # Software Foundation; either version 3 of the License, or (at your option) any
|
|
|
16 # later version.
|
|
|
17 #
|
|
|
18 # MayaChemTools is distributed in the hope that it will be useful, but without
|
|
|
19 # any warranty; without even the implied warranty of merchantability of fitness
|
|
|
20 # for a particular purpose. See the GNU Lesser General Public License for more
|
|
|
21 # details.
|
|
|
22 #
|
|
|
23 # You should have received a copy of the GNU Lesser General Public License
|
|
|
24 # along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or
|
|
|
25 # write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,
|
|
|
26 # Boston, MA, 02111-1307, USA.
|
|
|
27 #
|
|
|
28
|
|
|
29 use strict;
|
|
|
30 use FindBin; use lib "$FindBin::Bin/../lib";
|
|
|
31 use Getopt::Long;
|
|
|
32 use File::Basename;
|
|
|
33 use Text::ParseWords;
|
|
|
34 use Benchmark;
|
|
|
35 use FileUtil;
|
|
|
36 use TextUtil;
|
|
|
37 use MolecularFormula;
|
|
|
38
|
|
|
39 my($ScriptName, %Options, $StartTime, $EndTime, $TotalTime);
|
|
|
40
|
|
|
41 # Autoflush STDOUT
|
|
|
42 $| = 1;
|
|
|
43
|
|
|
44 # Starting message...
|
|
|
45 $ScriptName = basename($0);
|
|
|
46 print "\n$ScriptName: Starting...\n\n";
|
|
|
47 $StartTime = new Benchmark;
|
|
|
48
|
|
|
49 # Get the options and setup script...
|
|
|
50 SetupScriptUsage();
|
|
|
51 if ($Options{help} || @ARGV < 1) {
|
|
|
52 die GetUsageFromPod("$FindBin::Bin/$ScriptName");
|
|
|
53 }
|
|
|
54
|
|
|
55 my(@TextFilesList);
|
|
|
56 @TextFilesList = ExpandFileNames(\@ARGV, "csv tsv");
|
|
|
57
|
|
|
58 # Process options...
|
|
|
59 my(%OptionsInfo);
|
|
|
60 print "Processing options...\n";
|
|
|
61 ProcessOptions();
|
|
|
62
|
|
|
63 print "Checking input text file(s)...\n";
|
|
|
64 my(%TextFilesInfo);
|
|
|
65 RetrieveTextFilesInfo();
|
|
|
66 RetrieveColumnsAndLabelsInfo();
|
|
|
67
|
|
|
68 # Generate output files...
|
|
|
69 my($FileIndex);
|
|
|
70 if (@TextFilesList > 1) {
|
|
|
71 print "\nProcessing text files...\n";
|
|
|
72 }
|
|
|
73 for $FileIndex (0 .. $#TextFilesList) {
|
|
|
74 if ($TextFilesInfo{FileOkay}[$FileIndex]) {
|
|
|
75 print "\nProcessing file $TextFilesList[$FileIndex]...\n";
|
|
|
76 PerformElementalAnalysis($FileIndex);
|
|
|
77 }
|
|
|
78 }
|
|
|
79 print "\n$ScriptName:Done...\n\n";
|
|
|
80
|
|
|
81 $EndTime = new Benchmark;
|
|
|
82 $TotalTime = timediff ($EndTime, $StartTime);
|
|
|
83 print "Total time: ", timestr($TotalTime), "\n";
|
|
|
84
|
|
|
85 ###############################################################################
|
|
|
86
|
|
|
87 # Perform elemental analysis...
|
|
|
88 sub PerformElementalAnalysis {
|
|
|
89 my($Index) = @_;
|
|
|
90 my($TextFile, $NewTextFile, $FormulaCol, $Line, $NewLine, $FormulaColValue, $InDelim, $ColNum, $Value, $Status, $ErrorMsg, @ColLabels, @LineWords, @ColNumsBeforeNew, @ColNumsAfterNew);
|
|
|
91
|
|
|
92 $TextFile = $TextFilesList[$Index];
|
|
|
93 $InDelim = $TextFilesInfo{InDelim}[$Index];
|
|
|
94 $NewTextFile = $TextFilesInfo{OutFile}[$Index];
|
|
|
95 $FormulaCol = $TextFilesInfo{FormulaColNum}[$Index];
|
|
|
96
|
|
|
97 @ColNumsBeforeNew = @{$TextFilesInfo{ColNumsBeforeNew}[$Index]};
|
|
|
98 @ColNumsAfterNew = @{$TextFilesInfo{ColNumsAfterNew}[$Index]};
|
|
|
99
|
|
|
100 print "Generating new Text file $NewTextFile...\n";
|
|
|
101 open NEWTEXTFILE, ">$NewTextFile" or die "Error: Couldn't open $NewTextFile: $! \n";
|
|
|
102 open TEXTFILE, "$TextFile" or die "Error: Can't open $TextFile: $! \n";
|
|
|
103
|
|
|
104 # Skip over column labels from old file...
|
|
|
105 $Line = GetTextLine(\*TEXTFILE);
|
|
|
106
|
|
|
107 # Add column lablels in new file...
|
|
|
108 @ColLabels = ();
|
|
|
109 for $ColNum (@ColNumsBeforeNew) {
|
|
|
110 push @ColLabels, $TextFilesInfo{ColLabels}[$Index][$ColNum];
|
|
|
111 }
|
|
|
112 for $Value (@{$OptionsInfo{SpecifiedCalculations}}) {
|
|
|
113 push @ColLabels, $TextFilesInfo{ValueLabelsMap}[$Index]{$Value};
|
|
|
114 }
|
|
|
115 for $ColNum (@ColNumsAfterNew) {
|
|
|
116 push @ColLabels, $TextFilesInfo{ColLabels}[$Index][$ColNum];
|
|
|
117 }
|
|
|
118 $NewLine = '';
|
|
|
119 $NewLine = JoinWords(\@ColLabels, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
|
|
|
120 print NEWTEXTFILE "$NewLine\n";
|
|
|
121
|
|
|
122 # Go over all rows...
|
|
|
123 my($LineCount, $ElementsRef, $ElementCompositionRef, $CalculationType, $CalculatedValue, @CalculatedValues);
|
|
|
124
|
|
|
125 $LineCount = 1;
|
|
|
126 TEXTLINE: while ($Line = GetTextLine(\*TEXTFILE)) {
|
|
|
127 @LineWords = quotewords($InDelim, 0, $Line);
|
|
|
128 $LineCount++;
|
|
|
129
|
|
|
130 @CalculatedValues = ();
|
|
|
131 for $Value (@{$OptionsInfo{SpecifiedCalculations}}) {
|
|
|
132 push @CalculatedValues, '';
|
|
|
133 }
|
|
|
134 if ($FormulaCol > @LineWords) {
|
|
|
135 $ErrorMsg = "Ignoring line $LineCount: Formula column $ColLabels[$FormulaCol] not found";
|
|
|
136 PrintErrorMsg($Line, $ErrorMsg);
|
|
|
137 ComposeAndWriteNewLine(\*NEWTEXTFILE, \@LineWords, \@ColNumsBeforeNew, \@ColNumsAfterNew, \@CalculatedValues);
|
|
|
138 next TEXTLINE;
|
|
|
139 }
|
|
|
140
|
|
|
141 # Make sure it's a valid molecular formula...
|
|
|
142 $FormulaColValue = $LineWords[$FormulaCol];
|
|
|
143 if ($OptionsInfo{CheckFormula}) {
|
|
|
144 ($Status, $ErrorMsg) = MolecularFormula::IsMolecularFormula($FormulaColValue);
|
|
|
145 if (!$Status) {
|
|
|
146 $ErrorMsg = "Ignoring line $LineCount: Formula column $ColLabels[$FormulaCol] value is not valid: $ErrorMsg";
|
|
|
147 PrintErrorMsg($Line, $ErrorMsg);
|
|
|
148 ComposeAndWriteNewLine(\*NEWTEXTFILE, \@LineWords, \@ColNumsBeforeNew, \@ColNumsAfterNew, \@CalculatedValues);
|
|
|
149 next TEXTLINE;
|
|
|
150 }
|
|
|
151 }
|
|
|
152
|
|
|
153 # Calculate appropriate values and write 'em out...
|
|
|
154 @CalculatedValues = ();
|
|
|
155 for $CalculationType (@{$OptionsInfo{SpecifiedCalculations}}) {
|
|
|
156 if ($CalculationType =~ /^ElementalAnalysis$/i) {
|
|
|
157 ($ElementsRef, $ElementCompositionRef) = MolecularFormula::CalculateElementalComposition($FormulaColValue);
|
|
|
158 $CalculatedValue = (defined($ElementsRef) && defined($ElementCompositionRef)) ? MolecularFormula::FormatCompositionInfomation($ElementsRef, $ElementCompositionRef, $OptionsInfo{Precision}) : '';
|
|
|
159 }
|
|
|
160 elsif ($CalculationType =~ /^MolecularWeight$/i) {
|
|
|
161 $CalculatedValue = MolecularFormula::CalculateMolecularWeight($FormulaColValue);
|
|
|
162 $CalculatedValue = (defined($CalculatedValue) && length($CalculatedValue)) ? (sprintf("%.$OptionsInfo{Precision}f", $CalculatedValue)) : "";
|
|
|
163 }
|
|
|
164 elsif ($CalculationType =~ /^ExactMass$/i) {
|
|
|
165 $CalculatedValue = MolecularFormula::CalculateExactMass($FormulaColValue);
|
|
|
166 $CalculatedValue = (defined($CalculatedValue) && length($CalculatedValue)) ? (sprintf("%.$OptionsInfo{Precision}f", $CalculatedValue)) : "";
|
|
|
167 }
|
|
|
168 else {
|
|
|
169 $CalculatedValue = '';
|
|
|
170 }
|
|
|
171 push @CalculatedValues, $CalculatedValue;
|
|
|
172 }
|
|
|
173
|
|
|
174 ComposeAndWriteNewLine(\*NEWTEXTFILE, \@LineWords, \@ColNumsBeforeNew, \@ColNumsAfterNew, \@CalculatedValues);
|
|
|
175 }
|
|
|
176 close NEWTEXTFILE;
|
|
|
177 close TEXTFILE;
|
|
|
178
|
|
|
179 }
|
|
|
180
|
|
|
181 # Write out new line using old and new calculated data...
|
|
|
182 sub ComposeAndWriteNewLine {
|
|
|
183 my($NewTextFileRef, $OldLineWordsRef, $ColNumsBeforeNewRef, $ColNumsAfterNewRef, $CalculatedValuesRef) = @_;
|
|
|
184 my($NewLine, $ColNum, $Value, @NewLineWords);
|
|
|
185
|
|
|
186 @NewLineWords = ();
|
|
|
187 for $ColNum (@{$ColNumsBeforeNewRef}) {
|
|
|
188 push @NewLineWords, $OldLineWordsRef->[$ColNum];
|
|
|
189 }
|
|
|
190 for $Value (@{$CalculatedValuesRef}) {
|
|
|
191 push @NewLineWords, $Value;
|
|
|
192 }
|
|
|
193 for $ColNum (@{$ColNumsAfterNewRef}) {
|
|
|
194 push @NewLineWords, $OldLineWordsRef->[$ColNum];
|
|
|
195 }
|
|
|
196 $NewLine = JoinWords(\@NewLineWords, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
|
|
|
197 print $NewTextFileRef "$NewLine\n";
|
|
|
198 }
|
|
|
199
|
|
|
200 # Print out error message...
|
|
|
201 sub PrintErrorMsg {
|
|
|
202 my($Line, $ErrorMsg) = @_;
|
|
|
203
|
|
|
204 if ($OptionsInfo{DetailLevel} >= 2 ) {
|
|
|
205 print "$ErrorMsg: $Line\n";
|
|
|
206 }
|
|
|
207 elsif ($OptionsInfo{DetailLevel} >= 1) {
|
|
|
208 print "$ErrorMsg\n";
|
|
|
209 }
|
|
|
210 }
|
|
|
211
|
|
|
212 # Process formula columns and other information...
|
|
|
213 sub RetrieveColumnsAndLabelsInfo {
|
|
|
214 RetrieveFormulaColumnsInfo();
|
|
|
215 RetrieveStartColumnsAndValueLabelsInfo();
|
|
|
216 }
|
|
|
217
|
|
|
218 # Make sure specified formula column are okay...
|
|
|
219 sub RetrieveFormulaColumnsInfo {
|
|
|
220 my($Index, $TextFile);
|
|
|
221
|
|
|
222 @{$TextFilesInfo{FormulaColNum}} = ();
|
|
|
223
|
|
|
224 FILELIST: for $Index (0 .. $#TextFilesList) {
|
|
|
225 $TextFile = $TextFilesList[$Index];
|
|
|
226
|
|
|
227 $TextFilesInfo{FormulaColNum}[$Index] = 0;
|
|
|
228
|
|
|
229 if ($TextFilesInfo{FileOkay}[$Index]) {
|
|
|
230 my($FormulaColNum, $FormulaColValid);
|
|
|
231
|
|
|
232 $FormulaColNum = 0;
|
|
|
233 $FormulaColValid = 0;
|
|
|
234 if ($OptionsInfo{SpecifiedFormulaCol}) {
|
|
|
235 if ($OptionsInfo{ColMode} =~ /^colnum$/i) {
|
|
|
236 if ($OptionsInfo{SpecifiedFormulaCol} <= $TextFilesInfo{ColCount}[$Index]) {
|
|
|
237 $FormulaColNum = $OptionsInfo{SpecifiedFormulaCol} - 1;
|
|
|
238 $FormulaColValid = 1;
|
|
|
239 }
|
|
|
240 }
|
|
|
241 else {
|
|
|
242 if (exists($TextFilesInfo{ColLabelToNumMap}[$Index]{$OptionsInfo{SpecifiedFormulaCol}})) {
|
|
|
243 $FormulaColNum = $TextFilesInfo{ColLabelToNumMap}[$Index]{$OptionsInfo{SpecifiedFormulaCol}};
|
|
|
244 $FormulaColValid = 1;
|
|
|
245 }
|
|
|
246 }
|
|
|
247 }
|
|
|
248 else {
|
|
|
249 # Grab the first column with the word Formula in its label...
|
|
|
250 my($ColLabel);
|
|
|
251 LABEL: for $ColLabel (@{$TextFilesInfo{ColLabels}[$Index]}) {
|
|
|
252 if ($ColLabel =~ /Formula/i) {
|
|
|
253 $FormulaColNum = $TextFilesInfo{ColLabelToNumMap}[$Index]{$ColLabel};
|
|
|
254 $FormulaColValid = 1;
|
|
|
255 last LABEL;
|
|
|
256 }
|
|
|
257 }
|
|
|
258 }
|
|
|
259 if ($FormulaColValid) {
|
|
|
260 $TextFilesInfo{FormulaColNum}[$Index] = $FormulaColNum;
|
|
|
261 }
|
|
|
262 else {
|
|
|
263 if ($OptionsInfo{SpecifiedFormulaCol}) {
|
|
|
264 warn "Warning: Ignoring file $TextFile: Formula column specified, $OptionsInfo{SpecifiedFormulaCol}, using \"f --formulacol\" option doesn't exist\n";
|
|
|
265 }
|
|
|
266 else {
|
|
|
267 warn "Warning: Ignoring file $TextFile: Column label containing the word Formula doesn't exist\n";
|
|
|
268 }
|
|
|
269 $TextFilesInfo{FileOkay}[$Index] = 0;
|
|
|
270 }
|
|
|
271 }
|
|
|
272 }
|
|
|
273 }
|
|
|
274
|
|
|
275 # Setup starting column number for adding calculated values and
|
|
|
276 # column lables to use for these values...
|
|
|
277 sub RetrieveStartColumnsAndValueLabelsInfo {
|
|
|
278 my($Index, $TextFile, $SpecifiedStartColNum, $StartColNum, $Label, $Value, $NewLabel, $Count, $BeforeStartColNum, $AfterStartColNum, $FirstColNum, $LastColNum, $ColNum, $Part1StartColNum, $Part1EndColNum, $Part2StartColNum, $Part2EndColNum, @Part1ColNums, @Part2ColNums);
|
|
|
279
|
|
|
280 # Start column number for inserting new values...
|
|
|
281 $SpecifiedStartColNum = "last";
|
|
|
282 if (defined($OptionsInfo{StartCol})) {
|
|
|
283 if (length($OptionsInfo{StartCol})) {
|
|
|
284 $SpecifiedStartColNum = $OptionsInfo{StartCol}
|
|
|
285 }
|
|
|
286 }
|
|
|
287
|
|
|
288 # Column labels for for new calculated values...
|
|
|
289 my(%NewValueLabels) = (ElementalAnalysis => 'ElementalAnalysis', MolecularWeight => 'MolecularWeight', ExactMass => 'ExactMass');
|
|
|
290 if (@{$OptionsInfo{SpecifiedValueLabels}}) {
|
|
|
291 for ($Index = 0; $Index < @{$OptionsInfo{SpecifiedValueLabels}}; $Index +=2) {
|
|
|
292 $Value = $OptionsInfo{SpecifiedValueLabels}[$Index];
|
|
|
293 $Label = $OptionsInfo{SpecifiedValueLabels}[$Index + 1];
|
|
|
294 if (exists $NewValueLabels{$Value}) {
|
|
|
295 $NewValueLabels{$Value} = $Label;
|
|
|
296 }
|
|
|
297 }
|
|
|
298 }
|
|
|
299
|
|
|
300 @{$TextFilesInfo{ColNumsBeforeNew}} = ();
|
|
|
301 @{$TextFilesInfo{ColNumsAfterNew}} = ();
|
|
|
302 @{$TextFilesInfo{ValueLabelsMap}} = ();
|
|
|
303
|
|
|
304 FILELIST: for $Index (0 .. $#TextFilesList) {
|
|
|
305 $TextFile = $TextFilesList[$Index];
|
|
|
306
|
|
|
307 @{$TextFilesInfo{ColNumsBeforeNew}[$Index]} = ();
|
|
|
308 @{$TextFilesInfo{ColNumsAfterNew}[$Index]} = ();
|
|
|
309 %{$TextFilesInfo{ValueLabelsMap}[$Index]} = ();
|
|
|
310
|
|
|
311 if (!$TextFilesInfo{FileOkay}[$Index]) {
|
|
|
312 next FILELIST;
|
|
|
313 }
|
|
|
314
|
|
|
315 if ($SpecifiedStartColNum !~ /^last$/i) {
|
|
|
316 if ($OptionsInfo{ColMode} =~ /^collabel$/i) {
|
|
|
317 if (exists($TextFilesInfo{ColLabelToNumMap}[$Index]{$SpecifiedStartColNum})) {
|
|
|
318 $StartColNum = $TextFilesInfo{ColLabelToNumMap}[$Index]{$SpecifiedStartColNum};
|
|
|
319 }
|
|
|
320 else {
|
|
|
321 die "Error: Invalid value $SpecifiedStartColNum specified using \"-s --startcol\" option: column name doesn't exist in $TextFile \n";
|
|
|
322 }
|
|
|
323 }
|
|
|
324 else {
|
|
|
325 if ($SpecifiedStartColNum > 0 && $SpecifiedStartColNum <= $TextFilesInfo{ColCount}[$Index]) {
|
|
|
326 $StartColNum -= 1;
|
|
|
327 }
|
|
|
328 else {
|
|
|
329 die "Error: Invalid value $SpecifiedStartColNum specified using \"-s --startcol\" option: column number doesn't exist in $TextFile \n";
|
|
|
330 }
|
|
|
331 }
|
|
|
332 }
|
|
|
333 else {
|
|
|
334 $StartColNum = $TextFilesInfo{ColCount}[$Index] - 1;
|
|
|
335 }
|
|
|
336 # Set up columns lists for before and after the addition of calculated column values
|
|
|
337 # for each text file...
|
|
|
338 my($BeforeStartColNum, $AfterStartColNum, $FirstColNum, $LastColNum, $ColNum, $Part1StartColNum, $Part1EndColNum, $Part2StartColNum, $Part2EndColNum, @Part1ColNums, @Part2ColNums);
|
|
|
339
|
|
|
340 $FirstColNum = 0; $LastColNum = $TextFilesInfo{ColCount}[$Index] - 1;
|
|
|
341
|
|
|
342 $BeforeStartColNum = $StartColNum - 1;
|
|
|
343 $AfterStartColNum = $StartColNum + 1;
|
|
|
344
|
|
|
345 if ($OptionsInfo{StartColMode} =~ /^after$/i) {
|
|
|
346 $Part1StartColNum = $FirstColNum; $Part1EndColNum = $StartColNum;
|
|
|
347 $Part2StartColNum = $AfterStartColNum; $Part2EndColNum = $LastColNum;
|
|
|
348 }
|
|
|
349 else {
|
|
|
350 $Part1StartColNum = $FirstColNum; $Part1EndColNum = $BeforeStartColNum;
|
|
|
351 $Part2StartColNum = $StartColNum; $Part2EndColNum = $LastColNum;
|
|
|
352 }
|
|
|
353 @Part1ColNums = (); @Part2ColNums = ();
|
|
|
354 for $ColNum (0 .. $TextFilesInfo{ColCount}[$Index]) {
|
|
|
355 if ($ColNum >= $Part1StartColNum && $ColNum <= $Part1EndColNum) {
|
|
|
356 push @Part1ColNums, $ColNum;
|
|
|
357 }
|
|
|
358 }
|
|
|
359 for $ColNum (0 .. $TextFilesInfo{ColCount}[$Index]) {
|
|
|
360 if ($ColNum >= $Part2StartColNum && $ColNum <= $Part2EndColNum) {
|
|
|
361 push @Part2ColNums, $ColNum;
|
|
|
362 }
|
|
|
363 }
|
|
|
364 push @{$TextFilesInfo{ColNumsBeforeNew}[$Index]}, @Part1ColNums;
|
|
|
365 push @{$TextFilesInfo{ColNumsAfterNew}[$Index]}, @Part2ColNums;
|
|
|
366
|
|
|
367 # Setup column labels for calculated values...
|
|
|
368 for $Value (keys %NewValueLabels) {
|
|
|
369 $Label = $NewValueLabels{$Value};
|
|
|
370
|
|
|
371 # Make sure it doesn't already exists...
|
|
|
372 $Count = 1;
|
|
|
373 $NewLabel = $Label;
|
|
|
374 while (exists $TextFilesInfo{ColLabelToNumMap}[$Index]{$NewLabel}) {
|
|
|
375 $Count++;
|
|
|
376 $NewLabel = $Label . $Count;
|
|
|
377 }
|
|
|
378 $TextFilesInfo{ValueLabelsMap}[$Index]{$Value} = $NewLabel;
|
|
|
379 }
|
|
|
380 }
|
|
|
381 }
|
|
|
382
|
|
|
383 # Retrieve information about input text files...
|
|
|
384 sub RetrieveTextFilesInfo {
|
|
|
385 my($Index, $TextFile, $FileDir, $FileName, $FileExt, $InDelim, $Line, @ColLabels, $OutFileRoot, $OutFile, $ColNum, $ColLabel);
|
|
|
386
|
|
|
387 %TextFilesInfo = ();
|
|
|
388
|
|
|
389 @{$TextFilesInfo{FileOkay}} = ();
|
|
|
390 @{$TextFilesInfo{ColCount}} = ();
|
|
|
391 @{$TextFilesInfo{ColLabels}} = ();
|
|
|
392 @{$TextFilesInfo{ColLabelToNumMap}} = ();
|
|
|
393 @{$TextFilesInfo{InDelim}} = ();
|
|
|
394 @{$TextFilesInfo{OutFile}} = ();
|
|
|
395
|
|
|
396 FILELIST: for $Index (0 .. $#TextFilesList) {
|
|
|
397 $TextFile = $TextFilesList[$Index];
|
|
|
398
|
|
|
399 $TextFilesInfo{FileOkay}[$Index] = 0;
|
|
|
400 $TextFilesInfo{ColCount}[$Index] = 0;
|
|
|
401 $TextFilesInfo{InDelim}[$Index] = "";
|
|
|
402 $TextFilesInfo{OutFile}[$Index] = "";
|
|
|
403
|
|
|
404 @{$TextFilesInfo{ColLabels}[$Index]} = ();
|
|
|
405 %{$TextFilesInfo{ColLabelToNumMap}[$Index]} = ();
|
|
|
406
|
|
|
407 if (!(-e $TextFile)) {
|
|
|
408 warn "Warning: Ignoring file $TextFile: It doesn't exist\n";
|
|
|
409 next FILELIST;
|
|
|
410 }
|
|
|
411 if (!CheckFileType($TextFile, "csv tsv")) {
|
|
|
412 warn "Warning: Ignoring file $TextFile: It's not a csv or tsv file\n";
|
|
|
413 next FILELIST;
|
|
|
414 }
|
|
|
415 ($FileDir, $FileName, $FileExt) = ParseFileName($TextFile);
|
|
|
416 if ($FileExt =~ /^tsv$/i) {
|
|
|
417 $InDelim = "\t";
|
|
|
418 }
|
|
|
419 else {
|
|
|
420 $InDelim = "\,";
|
|
|
421 if ($Options{indelim} !~ /^(comma|semicolon)$/i) {
|
|
|
422 warn "Warning: Ignoring file $TextFile: The value specified, $Options{indelim}, for option \"--indelim\" is not valid for csv files\n";
|
|
|
423 next FILELIST;
|
|
|
424 }
|
|
|
425 if ($Options{indelim} =~ /^semicolon$/i) {
|
|
|
426 $InDelim = "\;";
|
|
|
427 }
|
|
|
428 }
|
|
|
429
|
|
|
430 if (!open TEXTFILE, "$TextFile") {
|
|
|
431 warn "Warning: Ignoring file $TextFile: Couldn't open it: $! \n";
|
|
|
432 next FILELIST;
|
|
|
433 }
|
|
|
434
|
|
|
435 $Line = GetTextLine(\*TEXTFILE);
|
|
|
436 @ColLabels = quotewords($InDelim, 0, $Line);
|
|
|
437 close TEXTFILE;
|
|
|
438
|
|
|
439 $FileDir = ""; $FileName = ""; $FileExt = "";
|
|
|
440 ($FileDir, $FileName, $FileExt) = ParseFileName($TextFile);
|
|
|
441 $FileExt = "csv";
|
|
|
442 if ($Options{outdelim} =~ /^tab$/i) {
|
|
|
443 $FileExt = "tsv";
|
|
|
444 }
|
|
|
445 if ($Options{root} && (@TextFilesList == 1)) {
|
|
|
446 my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($Options{root});
|
|
|
447 if ($RootFileName && $RootFileExt) {
|
|
|
448 $FileName = $RootFileName;
|
|
|
449 }
|
|
|
450 else {
|
|
|
451 $FileName = $Options{root};
|
|
|
452 }
|
|
|
453 $OutFileRoot = $FileName;
|
|
|
454 }
|
|
|
455 else {
|
|
|
456 $OutFileRoot = $FileName . "ElementalAnalysis";
|
|
|
457 }
|
|
|
458
|
|
|
459 $OutFile = $OutFileRoot . ".$FileExt";
|
|
|
460 if (lc($OutFile) eq lc($TextFile)) {
|
|
|
461 warn "Warning: Ignoring file $TextFile:Output file name, $OutFile, is same as input text file name, $TextFile\n";
|
|
|
462 next FILELIST;
|
|
|
463 }
|
|
|
464 if (!$Options{overwrite}) {
|
|
|
465 if (-e $OutFile) {
|
|
|
466 warn "Warning: Ignoring file $TextFile: The file $OutFile already exists\n";
|
|
|
467 next FILELIST;
|
|
|
468 }
|
|
|
469 }
|
|
|
470
|
|
|
471 $TextFilesInfo{FileOkay}[$Index] = 1;
|
|
|
472 $TextFilesInfo{InDelim}[$Index] = $InDelim;
|
|
|
473 $TextFilesInfo{OutFile}[$Index] = "$OutFile";
|
|
|
474
|
|
|
475 $TextFilesInfo{ColCount}[$Index] = @ColLabels;
|
|
|
476 push @{$TextFilesInfo{ColLabels}[$Index]}, @ColLabels;
|
|
|
477 for $ColNum (0 .. $#ColLabels) {
|
|
|
478 $ColLabel = $ColLabels[$ColNum];
|
|
|
479 $TextFilesInfo{ColLabelToNumMap}[$Index]{$ColLabel} = $ColNum;
|
|
|
480 }
|
|
|
481 }
|
|
|
482
|
|
|
483 }
|
|
|
484
|
|
|
485 # Process option values...
|
|
|
486 sub ProcessOptions {
|
|
|
487 %OptionsInfo = ();
|
|
|
488
|
|
|
489 $OptionsInfo{Mode} = $Options{mode};
|
|
|
490
|
|
|
491 $OptionsInfo{ColMode} = $Options{colmode};
|
|
|
492 $OptionsInfo{StartColMode} = $Options{startcolmode};
|
|
|
493
|
|
|
494 $OptionsInfo{Fast} = defined $Options{fast} ? $Options{fast} : undef;
|
|
|
495
|
|
|
496 $OptionsInfo{DetailLevel} = $Options{detail};
|
|
|
497 $OptionsInfo{CheckFormula} = $Options{fast} ? 0 : 1;
|
|
|
498 $OptionsInfo{Precision} = $Options{precision};
|
|
|
499
|
|
|
500 $OptionsInfo{InDelim} = $Options{indelim};
|
|
|
501
|
|
|
502 $OptionsInfo{OutDelim} = ($Options{outdelim} =~ /^tab$/i ) ? "\t" : (($Options{outdelim} =~ /^semicolon$/i) ? "\;" : "\,");
|
|
|
503 $OptionsInfo{OutQuote} = ($Options{quote} =~ /^yes$/i) ? 1 : 0;
|
|
|
504
|
|
|
505 $OptionsInfo{Overwrite} = defined $Options{overwrite} ? $Options{overwrite} : undef;
|
|
|
506 $OptionsInfo{OutFileRoot} = defined $Options{root} ? $Options{root} : undef;
|
|
|
507
|
|
|
508 $OptionsInfo{StartCol} = defined $Options{startcol} ? $Options{startcol} : undef;
|
|
|
509
|
|
|
510 $OptionsInfo{FormulaCol} = defined $Options{formulacol} ? $Options{formulacol} : undef;
|
|
|
511 $OptionsInfo{SpecifiedFormulaCol} = "";
|
|
|
512
|
|
|
513 if (defined $Options{formulacol}) {
|
|
|
514 $OptionsInfo{SpecifiedFormulaCol} = $Options{formulacol};
|
|
|
515 if ($Options{colmode} =~ /^colnum$/i) {
|
|
|
516 if (!IsPositiveInteger($OptionsInfo{SpecifiedFormulaCol})) {
|
|
|
517 die "Error: Invalid value $Options{formulacol} specified using \"-f -formulacol\" option: Allowed values: > 0\n";
|
|
|
518 }
|
|
|
519 }
|
|
|
520 }
|
|
|
521
|
|
|
522 # Setup what to calculate...
|
|
|
523 @{$OptionsInfo{SpecifiedCalculations}} = ();
|
|
|
524 if ($Options{mode} =~ /^All$/i) {
|
|
|
525 @{$OptionsInfo{SpecifiedCalculations}} = qw(ElementalAnalysis MolecularWeight ExactMass);
|
|
|
526 }
|
|
|
527 else {
|
|
|
528 my($Mode, $ModeValue, @SpecifiedModeValues);
|
|
|
529 $Mode = $Options{mode};
|
|
|
530 $Mode =~ s/ //g;
|
|
|
531 @SpecifiedModeValues = split /\,/, $Mode;
|
|
|
532 for $ModeValue (@SpecifiedModeValues) {
|
|
|
533 if ($ModeValue !~ /^(ElementalAnalysis|MolecularWeight|ExactMass)$/i) {
|
|
|
534 if ($ModeValue =~ /^All$/i) {
|
|
|
535 die "Error: All value for option \"-m --mode\" is not allowed with other valid values.\n";
|
|
|
536 }
|
|
|
537 else {
|
|
|
538 die "Error: The value specified, $ModeValue, for option \"-m --mode\" is not valid. Allowed values: ElementalAnalysis, MolecularWeight, or ExactMass\n";
|
|
|
539 }
|
|
|
540 }
|
|
|
541 push @{$OptionsInfo{SpecifiedCalculations}}, $ModeValue;
|
|
|
542 }
|
|
|
543 }
|
|
|
544
|
|
|
545 $OptionsInfo{ValueColLabels} = defined $Options{valuecollabels} ? $Options{valuecollabels} : undef;
|
|
|
546 @{$OptionsInfo{SpecifiedValueLabels}} = ();
|
|
|
547
|
|
|
548 if ($Options{valuecollabels}) {
|
|
|
549 my($Value, $Label, @ValueLabels);
|
|
|
550 @ValueLabels = split /\,/, $Options{valuecollabels};
|
|
|
551 if (@ValueLabels % 2) {
|
|
|
552 die "Error: The value specified, $Options{valuecollabels}, for option \"-v --valuecollabels\" is not valid: It must contain even number of comma delimited values\n";
|
|
|
553 }
|
|
|
554 my($Index);
|
|
|
555 for ($Index = 0; $Index < @ValueLabels; $Index +=2) {
|
|
|
556 $Value = $ValueLabels[$Index];
|
|
|
557 $Value =~ s/ //g;
|
|
|
558 $Label = $ValueLabels[$Index + 1];
|
|
|
559 if ($Value !~ /^(ElementalAnalysis|MolecularWeight|ExactMass)$/i) {
|
|
|
560 die "Error: The value specified, $Value, using option \"-v --valuecollabels\" is not valid. Allowed values: ElementalAnalysis, MolecularWeight, or ExactMass\n";
|
|
|
561 }
|
|
|
562 push @{$OptionsInfo{SpecifiedValueLabels}}, ($Value, $Label);
|
|
|
563 }
|
|
|
564 }
|
|
|
565 }
|
|
|
566
|
|
|
567 # Setup script usage and retrieve command line arguments specified using various options...
|
|
|
568 sub SetupScriptUsage {
|
|
|
569
|
|
|
570 # Retrieve all the options...
|
|
|
571 %Options = ();
|
|
|
572 $Options{colmode} = "colnum";
|
|
|
573 $Options{detail} = 1;
|
|
|
574 $Options{mode} = "All";
|
|
|
575 $Options{indelim} = "comma";
|
|
|
576 $Options{outdelim} = "comma";
|
|
|
577 $Options{precision} = 2;
|
|
|
578 $Options{quote} = "yes";
|
|
|
579 $Options{startcolmode} = "after";
|
|
|
580
|
|
|
581 if (!GetOptions(\%Options, "colmode|c=s", "detail|d=i", "fast", "formulacol|f=s", "help|h", "indelim=s", "mode|m=s", "outdelim=s", "overwrite|o", "precision|p=i", "quote|q=s", "root|r=s", "startcol|s=s", "startcolmode=s", "valuecollabels|v=s", "workingdir|w=s")) {
|
|
|
582 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
|
|
|
583 }
|
|
|
584 if ($Options{workingdir}) {
|
|
|
585 if (! -d $Options{workingdir}) {
|
|
|
586 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n";
|
|
|
587 }
|
|
|
588 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
|
|
|
589 }
|
|
|
590 if ($Options{colmode} !~ /^(colnum|collabel)$/i) {
|
|
|
591 die "Error: The value specified, $Options{colmode}, for option \"-c --colmode\" is not valid. Allowed values: colnum or collabel\n";
|
|
|
592 }
|
|
|
593 if ($Options{indelim} !~ /^(comma|semicolon)$/i) {
|
|
|
594 die "Error: The value specified, $Options{indelim}, for option \"--indelim\" is not valid. Allowed values: comma or semicolon\n";
|
|
|
595 }
|
|
|
596 if ($Options{outdelim} !~ /^(comma|semicolon|tab)$/i) {
|
|
|
597 die "Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n";
|
|
|
598 }
|
|
|
599 if (!IsPositiveInteger($Options{precision})) {
|
|
|
600 die "Error: The value specified, $Options{precision}, for option \"-p --precision\" is not valid. Allowed values: > 0 \n";
|
|
|
601 }
|
|
|
602 if ($Options{quote} !~ /^(yes|no)$/i) {
|
|
|
603 die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n";
|
|
|
604 }
|
|
|
605 if ($Options{startcolmode} !~ /^(before|after)$/i) {
|
|
|
606 die "Error: The value specified, $Options{quote}, for option \"--startcolmode\" is not valid. Allowed values: before or after\n";
|
|
|
607 }
|
|
|
608 if (!IsPositiveInteger($Options{detail})) {
|
|
|
609 die "Error: The value specified, $Options{detail}, for option \"-d --detail\" is not valid. Allowed values: > 0\n";
|
|
|
610 }
|
|
|
611 }
|
|
|
612
|
|
|
613 __END__
|
|
|
614
|
|
|
615 =head1 NAME
|
|
|
616
|
|
|
617 ElementalAnalysisTextFiles.pl - Perform elemental analysis using formula column in TextFile(s)
|
|
|
618
|
|
|
619 =head1 SYNOPSIS
|
|
|
620
|
|
|
621 ElementalAnalysisTextFiles.pl TextFile(s)...
|
|
|
622
|
|
|
623 ElementalAnalysisTextFiles.pl [B<-c, --colmode> colnum | collabel] [B<-d, --detail> infolevel] [B<-f, --fast>]
|
|
|
624 [B<-f, --formulacol> colnum | collabel] [B<-h, --help>] [B<--indelim> comma | semicolon]
|
|
|
625 [B<-m, --mode> All | "ElementalAnysis, [MolecularWeight, ExactMass]"] [B<-o, --overwrite>]
|
|
|
626 [B<--outdelim> comma | tab | semicolon] [B<-p, --precision> number] [B<-q, --quote> yes | no]
|
|
|
627 [B<-r, --root> rootname] [B<-s, --startcol> colnum | collabel] [B<--startcolmode> before | after]
|
|
|
628 B<-v --valuecollabels> [Name, Label, [Name, Label,...]] [B<-w, --workingdir> dirname] TextFile(s)...
|
|
|
629
|
|
|
630 =head1 DESCRIPTION
|
|
|
631
|
|
|
632 Perform elemental analysis using molecular formula column specified by a column number or label in
|
|
|
633 I<TextFile(s)>.
|
|
|
634
|
|
|
635 In addition to straightforward molecular formulas - H2O, HCl, C3H7O2N -
|
|
|
636 other supported variations are: Ca3(PO4)2, [PCl4]+, [Fe(CN)6]4-, C37H42N2O6+2, Na2CO3.10H2O,
|
|
|
637 8H2S.46H2O, and so on. Charges are simply ignored. Isotope symbols in formulas specification, including
|
|
|
638 D and T, are not supported.
|
|
|
639
|
|
|
640 The valid file extensions are I<.csv> and I<.tsv> for comma/semicolon and tab delimited
|
|
|
641 text files respectively. All other file names are ignored. All the text files in a
|
|
|
642 current directory can be specified by I<*.csv>, I<*.tsv>, or the current directory
|
|
|
643 name. The B<--indelim> option determines the format of I<TextFile(s)>. Any file
|
|
|
644 which doesn't correspond to the format indicated by B<--indelim> option is ignored.
|
|
|
645
|
|
|
646 =head1 OPTIONS
|
|
|
647
|
|
|
648 =over 4
|
|
|
649
|
|
|
650 =item B<-c, --colmode> I<colnum | collabel>
|
|
|
651
|
|
|
652 Specify how columns are identified in I<TextFile(s)>: using column number or column
|
|
|
653 label. Possible values: I<colnum or collabel>. Default value: I<colnum>.
|
|
|
654
|
|
|
655 =item B<-d, --detail> I<infolevel>
|
|
|
656
|
|
|
657 Level of information to print about lines being ignored. Default: I<1>. Possible values:
|
|
|
658 I<1, 2 or 3>.
|
|
|
659
|
|
|
660 =item B<-h, --help>
|
|
|
661
|
|
|
662 Print this help message.
|
|
|
663
|
|
|
664 =item B<--fast>
|
|
|
665
|
|
|
666 In this mode, the formula column specified using B<-f, --formulacol> option is assumed
|
|
|
667 to contain valid molecular formula data and initial formula validation check is skipped.
|
|
|
668
|
|
|
669 =item B<-f, --formulacol> I<col number | col name>
|
|
|
670
|
|
|
671 This value is mode specific. It specifies molecular formula column to use for performing
|
|
|
672 elemental analysis on I<TextFile(s)>. Possible values: I<col number or col label>.
|
|
|
673 Default value: I<first column containing the word formula in its column label>.
|
|
|
674
|
|
|
675 =item B<-m, --mode> I<All | "ElementalAnalysis,[MolecularWeight,ExactMass]">
|
|
|
676
|
|
|
677 Specify what values to calculate using molecular formula in I<TextFile(s)>: calculate all supported
|
|
|
678 values or specify a comma delimited list of values. Possible values: I<All | "ElementalAnalysis,
|
|
|
679 [MolecularWeight, ExactMass]">. Default: I<All>
|
|
|
680
|
|
|
681 =item B<--indelim> I<comma | semicolon>
|
|
|
682
|
|
|
683 Input delimiter for CSV I<TextFile(s)>. Possible values: I<comma or semicolon>.
|
|
|
684 Default value: I<comma>. For TSV files, this option is ignored and I<tab> is used as a
|
|
|
685 delimiter.
|
|
|
686
|
|
|
687 =item B<-o, --overwrite>
|
|
|
688
|
|
|
689 Overwrite existing files.
|
|
|
690
|
|
|
691 =item B<--outdelim> I<comma | tab | semicolon>
|
|
|
692
|
|
|
693 Output text file delimiter. Possible values: I<comma, tab, or semicolon>
|
|
|
694 Default value: I<comma>.
|
|
|
695
|
|
|
696 =item B<-p, --precision> I<number>
|
|
|
697
|
|
|
698 Precision of calculated values in the output file. Default: up to I<2> decimal places.
|
|
|
699 Valid values: positive integers.
|
|
|
700
|
|
|
701 =item B<-q, --quote> I<yes | no>
|
|
|
702
|
|
|
703 Put quotes around column values in output text file. Possible values: I<yes or
|
|
|
704 no>. Default value: I<yes>.
|
|
|
705
|
|
|
706 =item B<-r, --root> I<rootname>
|
|
|
707
|
|
|
708 New text file name is generated using the root: <Root>.<Ext>. Default new file
|
|
|
709 name: <InitialTextFileName>ElementalAnalysis.<Ext>. The csv, and tsv
|
|
|
710 <Ext> values are used for comma/semicolon, and tab delimited text files
|
|
|
711 respectively. This option is ignored for multiple input files.
|
|
|
712
|
|
|
713 =item B<-s, --startcol> I<colnum | collabel>
|
|
|
714
|
|
|
715 This value is mode specific. It specifies the column in text files which is
|
|
|
716 used for start adding calculated column values. For I<colnum> mode, specify
|
|
|
717 column number and for I<collabel> mode, specify column label.
|
|
|
718
|
|
|
719 Default value: I<last>. Start merge after the last column.
|
|
|
720
|
|
|
721 =item B<--startcolmode> I<before | after>
|
|
|
722
|
|
|
723 Start adding calculated column values after the B<-s, --startcol> value. Possible values: I<before or after>.
|
|
|
724 Default value: I<after>.
|
|
|
725
|
|
|
726 =item B<-v --valuecollabels> I<Name,Label,[Name,Label,...]>
|
|
|
727
|
|
|
728 Specify column labels to use for calculated values. In general, it's a comma delimited
|
|
|
729 list of value name and column label pairs. Supported value names: I<ElementalAnalysis,
|
|
|
730 MolecularWeight, and ExactMass>. Default labels: I<ElementalAnalysis, MolecularWeight,
|
|
|
731 and ExactMass>.
|
|
|
732
|
|
|
733 =item B<-w, --workingdir> I<dirname>
|
|
|
734
|
|
|
735 Location of working directory. Default: current directory.
|
|
|
736
|
|
|
737 =back
|
|
|
738
|
|
|
739 =head1 EXAMPLES
|
|
|
740
|
|
|
741 To perform elemental analysis, calculate molecular weight and exact mass using formulas
|
|
|
742 in a column with the word Formula in its column label and generate a new CSV text
|
|
|
743 file NewSample1.csv, type:
|
|
|
744
|
|
|
745 % ElementalAnalysisTextFiles.pl -o -r NewSample1 Sample1.csv
|
|
|
746
|
|
|
747 To perform elemental analysis using formulas in column number two, use column label
|
|
|
748 Analysis for calculated data, and generate a new CSV text file NewSample1.csv, type:
|
|
|
749
|
|
|
750 % ElementalAnalysisTextFiles.pl --m ElementalAnalysis --formulacol 2
|
|
|
751 --valuecollabels "ElementalAnalysis,Analysis" -o -r NewSample1
|
|
|
752 Sample1.csv
|
|
|
753
|
|
|
754 To calculate molecular weight using formula in column label Formula with four decimal
|
|
|
755 precision and generate a new CSV text file NewSample1.csv, type
|
|
|
756
|
|
|
757 % ElementalAnalysisTextFiles.pl --m MolecularWeight --colmode collabel
|
|
|
758 --formulacol Formula --precision 4 -o -r NewSample1 Sample1.csv
|
|
|
759
|
|
|
760 To calculate exact mass using formula in column label Formula with four decimal
|
|
|
761 precision, adding column for exact mass right after Formula column, and generate a
|
|
|
762 new CSV text file NewSample1.csv, type
|
|
|
763
|
|
|
764 % ElementalAnalysisTextFiles.pl --m ExactMass --colmode collabel
|
|
|
765 --formulacol Formula --precision 4 --startcolmode after
|
|
|
766 --startcol Formula -o -r NewSample1 Sample1.csv
|
|
|
767
|
|
|
768
|
|
|
769 =head1 AUTHOR
|
|
|
770
|
|
|
771 Manish Sud <msud@san.rr.com>
|
|
|
772
|
|
|
773 =head1 SEE ALSO
|
|
|
774
|
|
|
775 AnalyzeTextFilesData.pl, InfoTextFiles.pl, ExtractFromTextFiles.pl
|
|
|
776
|
|
|
777 =head1 COPYRIGHT
|
|
|
778
|
|
|
779 Copyright (C) 2015 Manish Sud. All rights reserved.
|
|
|
780
|
|
|
781 This file is part of MayaChemTools.
|
|
|
782
|
|
|
783 MayaChemTools is free software; you can redistribute it and/or modify it under
|
|
|
784 the terms of the GNU Lesser General Public License as published by the Free
|
|
|
785 Software Foundation; either version 3 of the License, or (at your option)
|
|
|
786 any later version.
|
|
|
787
|
|
|
788 =cut
|
