|
0
|
1 #!/usr/bin/perl -w
|
|
|
2 #
|
|
|
3 # $RCSfile: InfoNucleicAcids.pl,v $
|
|
|
4 # $Date: 2015/02/28 20:46:20 $
|
|
|
5 # $Revision: 1.26 $
|
|
|
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 "$FindBin::Bin/../lib";
|
|
|
31 use Getopt::Long;
|
|
|
32 use File::Basename;
|
|
|
33 use Text::ParseWords;
|
|
|
34 use Benchmark;
|
|
|
35 use FileUtil;
|
|
|
36 use TextUtil;
|
|
|
37 use NucleicAcids;
|
|
|
38
|
|
|
39 my($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}) {
|
|
|
52 die GetUsageFromPod("$FindBin::Bin/$ScriptName");
|
|
|
53 }
|
|
|
54
|
|
|
55 print "Processing options...\n";
|
|
|
56 my(%OptionsInfo);
|
|
|
57 ProcessOptions();
|
|
|
58
|
|
|
59 ListNucleicAcidProperties();
|
|
|
60 print "\n$ScriptName:Done...\n\n";
|
|
|
61
|
|
|
62 $EndTime = new Benchmark;
|
|
|
63 $TotalTime = timediff ($EndTime, $StartTime);
|
|
|
64 print "Total time: ", timestr($TotalTime), "\n";
|
|
|
65
|
|
|
66 ###############################################################################
|
|
|
67
|
|
|
68 # List data for an nucleic acid...
|
|
|
69 sub ListNucleicAcidData {
|
|
|
70 my($DataLabelRef, $DataValueRef) = @_;
|
|
|
71 my($Index, $Line, $Value);
|
|
|
72
|
|
|
73 if ($OptionsInfo{NucleicAcidRowsOutput}) {
|
|
|
74 $Line = '';
|
|
|
75 # Format data...
|
|
|
76 if ($OptionsInfo{OutQuote} || $Options{outdelim} !~ /^comma$/i) {
|
|
|
77 $Line = JoinWords($DataValueRef, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
|
|
|
78 }
|
|
|
79 else {
|
|
|
80 # Always quote values containing commas...
|
|
|
81 $Line = ($DataValueRef->[0] =~ /\,/) ? qq("$DataValueRef->[0]") : $DataValueRef->[0];
|
|
|
82 for $Index (1 .. $#{$DataValueRef} ) {
|
|
|
83 $Value = $DataValueRef->[$Index];
|
|
|
84 if ($Value =~ /\,/) {
|
|
|
85 $Value = qq("$Value");
|
|
|
86 }
|
|
|
87 $Line .= $OptionsInfo{OutDelim} . $Value;
|
|
|
88 }
|
|
|
89 }
|
|
|
90 if ($OptionsInfo{FileOutput}) {
|
|
|
91 print OUTFILE "$Line\n";
|
|
|
92 }
|
|
|
93 else {
|
|
|
94 print "$Line\n";
|
|
|
95 }
|
|
|
96 }
|
|
|
97 else {
|
|
|
98 # Format and list data...
|
|
|
99 $Line = '';
|
|
|
100 for $Index (0 .. $#{$DataLabelRef} ) {
|
|
|
101 $Line = $DataLabelRef->[$Index] . ' ' . $DataValueRef->[$Index];
|
|
|
102 if ($OptionsInfo{FileOutput}) {
|
|
|
103 print OUTFILE "$Line\n";
|
|
|
104 }
|
|
|
105 else {
|
|
|
106 print "$Line\n";
|
|
|
107 }
|
|
|
108 }
|
|
|
109 }
|
|
|
110 }
|
|
|
111
|
|
|
112 # List data for an nucleic acid...
|
|
|
113 sub ListHeaderRowData {
|
|
|
114 my($DataLabelRef) = @_;
|
|
|
115 my($Line);
|
|
|
116
|
|
|
117 # Format data...
|
|
|
118 $Line = JoinWords($DataLabelRef, $OptionsInfo{OutDelim}, $OptionsInfo{OutQuote});
|
|
|
119 $Line =~ s/\://g;
|
|
|
120 # List data...
|
|
|
121 if ($OptionsInfo{FileOutput}) {
|
|
|
122 print OUTFILE "$Line\n";
|
|
|
123 }
|
|
|
124 else {
|
|
|
125 print "$Line\n";
|
|
|
126 }
|
|
|
127 }
|
|
|
128
|
|
|
129 # List properties for nucleic acids...
|
|
|
130 sub ListNucleicAcidProperties {
|
|
|
131 my($NucleicAcidID, $NucleicAcidDataRef, $PropertyName, $PropertyValue, @PropertyLabels, @PropertyValues);
|
|
|
132
|
|
|
133 print "Listing information for nucleic acid(s)...\n";
|
|
|
134
|
|
|
135 if ($OptionsInfo{FileOutput}) {
|
|
|
136 print "Generating file $OptionsInfo{OutFileName}...\n";
|
|
|
137 open OUTFILE, ">$OptionsInfo{OutFileName}" or die "Couldn't open $OptionsInfo{OutFileName}: $!\n";
|
|
|
138 }
|
|
|
139
|
|
|
140 # Setup property labels...
|
|
|
141 @PropertyLabels = ();
|
|
|
142 for $PropertyName (@{$OptionsInfo{SpecifiedProperies}}) {
|
|
|
143 push @PropertyLabels, ("$PropertyName:");
|
|
|
144 }
|
|
|
145
|
|
|
146 if ($OptionsInfo{NucleicAcidRowsOutput}) {
|
|
|
147 ListHeaderRowData(\@PropertyLabels);
|
|
|
148 }
|
|
|
149
|
|
|
150 # Go over specified properties...
|
|
|
151 for $NucleicAcidID (@{$OptionsInfo{SpecifiedNucleicAcidIDs}}) {
|
|
|
152 $NucleicAcidDataRef = NucleicAcids::GetNucleicAcidPropertiesData($NucleicAcidID);
|
|
|
153
|
|
|
154 if (!$OptionsInfo{NucleicAcidRowsOutput}) {
|
|
|
155 if ($OptionsInfo{FileOutput}) {
|
|
|
156 print OUTFILE "\nListing properties for nucleic acid $NucleicAcidID...\n\n";
|
|
|
157 }
|
|
|
158 else {
|
|
|
159 print "\nListing properties for nucleic acid $NucleicAcidID...\n\n";
|
|
|
160 }
|
|
|
161 }
|
|
|
162
|
|
|
163 # Collect data..
|
|
|
164 @PropertyValues = ();
|
|
|
165 for $PropertyName (@{$OptionsInfo{SpecifiedProperies}}) {
|
|
|
166 $PropertyValue = $NucleicAcidDataRef->{$PropertyName};
|
|
|
167 if (IsFloat($PropertyValue)) {
|
|
|
168 $PropertyValue = sprintf("%.$OptionsInfo{Precision}f", $PropertyValue) + 0;
|
|
|
169 }
|
|
|
170 push @PropertyValues, $PropertyValue;
|
|
|
171 }
|
|
|
172 # List data...
|
|
|
173 ListNucleicAcidData(\@PropertyLabels, \@PropertyValues);
|
|
|
174 }
|
|
|
175 if ($OptionsInfo{FileOutput}) {
|
|
|
176 close OUTFILE;
|
|
|
177 }
|
|
|
178 print "\n";
|
|
|
179 }
|
|
|
180
|
|
|
181 # Get propery names from categories...
|
|
|
182 sub GetPropertyNamesFromCategories {
|
|
|
183 my($CategoryName) = @_;
|
|
|
184 my(@PropertyNames);
|
|
|
185
|
|
|
186 @PropertyNames = ();
|
|
|
187 if ($CategoryName =~ /^Basic$/i) {
|
|
|
188 @PropertyNames = ('Code', 'OtherCodes', 'Name', 'Type', 'MolecularFormula', 'MolecularWeight');
|
|
|
189 } elsif ($CategoryName =~ /^BasicPlus$/i) {
|
|
|
190 @PropertyNames = ('Code', 'OtherCodes', 'Name', 'Type', 'MolecularFormula', 'MolecularWeight', 'ExactMass', 'ElementalComposition');
|
|
|
191 }
|
|
|
192
|
|
|
193 return @PropertyNames;
|
|
|
194 }
|
|
|
195
|
|
|
196 # Process option values...
|
|
|
197 sub ProcessOptions {
|
|
|
198 %OptionsInfo = ();
|
|
|
199
|
|
|
200 $OptionsInfo{Mode} = $Options{mode};
|
|
|
201
|
|
|
202 $OptionsInfo{OutDelim} = ($Options{outdelim} =~ /^tab$/i ) ? "\t" : (($Options{outdelim} =~ /^semicolon$/i) ? "\;" : "\,");
|
|
|
203 $OptionsInfo{OutQuote} = ($Options{quote} =~ /^yes$/i) ? 1 : 0;
|
|
|
204
|
|
|
205 $OptionsInfo{Overwrite} = defined $Options{overwrite} ? $Options{overwrite} : undef;
|
|
|
206 $OptionsInfo{OutFileRoot} = defined $Options{root} ? $Options{root} : undef;
|
|
|
207
|
|
|
208 $OptionsInfo{Output} = $Options{output};
|
|
|
209 $OptionsInfo{OutputStyle} = $Options{outputstyle};
|
|
|
210
|
|
|
211 $OptionsInfo{NucleicAcidRowsOutput} = ($Options{outputstyle} =~ /^NucleicAcidRows$/i) ? 1 : 0;
|
|
|
212 $OptionsInfo{FileOutput} = ($Options{output} =~ /^File$/i) ? 1 : 0;
|
|
|
213
|
|
|
214 $OptionsInfo{Precision} = $Options{precision};
|
|
|
215
|
|
|
216 my($NucleicAcidID, @NucleicAcidIDs);
|
|
|
217
|
|
|
218 @{$OptionsInfo{SpecifiedNucleicAcidIDs}} = ();
|
|
|
219
|
|
|
220 # Set up Nucleic Acids IDs except for All mode...
|
|
|
221 @NucleicAcidIDs = ();
|
|
|
222
|
|
|
223 if (@ARGV >= 1) {
|
|
|
224 push @NucleicAcidIDs, @ARGV;
|
|
|
225 }
|
|
|
226 else {
|
|
|
227 # Setup mode specified default values...
|
|
|
228 if ($Options{mode} =~ /NucleicAcidID/i) {
|
|
|
229 push @NucleicAcidIDs, 'A';
|
|
|
230 }
|
|
|
231 elsif ($Options{mode} =~ /NucleicAcidType/i) {
|
|
|
232 push @NucleicAcidIDs, 'Nucleoside';
|
|
|
233 }
|
|
|
234 else {
|
|
|
235 push @NucleicAcidIDs, 'A';
|
|
|
236 }
|
|
|
237 }
|
|
|
238
|
|
|
239 # Generate list of nucleic acids...
|
|
|
240 if (@ARGV == 1 && $ARGV[0] =~ /^All$/i) {
|
|
|
241 push @{$OptionsInfo{SpecifiedNucleicAcidIDs}}, NucleicAcids::GetNucleicAcids();
|
|
|
242 }
|
|
|
243 else {
|
|
|
244 if ($Options{mode} =~ /NucleicAcidID/i) {
|
|
|
245 ID: for $NucleicAcidID (@NucleicAcidIDs) {
|
|
|
246 if (NucleicAcids::IsNucleicAcid($NucleicAcidID)) {
|
|
|
247 push @{$OptionsInfo{SpecifiedNucleicAcidIDs}}, $NucleicAcidID;
|
|
|
248 }
|
|
|
249 else {
|
|
|
250 warn "Ignoring nucleic acid ID, $NucleicAcidID, specified using command line parameter option: Unknown nucleic acid ID...\n";
|
|
|
251 next ID;
|
|
|
252 }
|
|
|
253 }
|
|
|
254 }
|
|
|
255 elsif ($Options{mode} =~ /NucleicAcidType/i) {
|
|
|
256 ID: for $NucleicAcidID (@NucleicAcidIDs) {
|
|
|
257 if (!NucleicAcids::IsNucleicAcidType($NucleicAcidID)) {
|
|
|
258 warn "Ignoring nucleic acid type, $NucleicAcidID, specified using command line parameter option: Unknown nucleic acid type...\n";
|
|
|
259 next ID;
|
|
|
260 }
|
|
|
261 push @{$OptionsInfo{SpecifiedNucleicAcidIDs}}, NucleicAcids::GetNucleicAcidsByType($NucleicAcidID);
|
|
|
262 }
|
|
|
263 }
|
|
|
264 }
|
|
|
265 SetupSpecifiedProperties();
|
|
|
266
|
|
|
267 # Setup output file name...
|
|
|
268 $OptionsInfo{OutFileName} = '';
|
|
|
269 if ($OptionsInfo{FileOutput}) {
|
|
|
270 my($OutFileRoot, $OutFileExt);
|
|
|
271
|
|
|
272 $OutFileRoot = '';
|
|
|
273 $OutFileExt = "csv";
|
|
|
274 if ($Options{outdelim} =~ /^tab$/i) {
|
|
|
275 $OutFileExt = "tsv";
|
|
|
276 }
|
|
|
277 if ($Options{root}) {
|
|
|
278 my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($Options{root});
|
|
|
279 if ($RootFileName && $RootFileExt) {
|
|
|
280 $OutFileRoot = $RootFileName;
|
|
|
281 }
|
|
|
282 else {
|
|
|
283 $OutFileRoot = $Options{root};
|
|
|
284 }
|
|
|
285 }
|
|
|
286 else {
|
|
|
287 $OutFileRoot = 'NucleicAcidsInfo';
|
|
|
288 }
|
|
|
289 $OptionsInfo{OutFileName} = $OutFileRoot . '.' . $OutFileExt;
|
|
|
290 if (!$Options{overwrite}) {
|
|
|
291 if (-e $OptionsInfo{OutFileName}) {
|
|
|
292 die "Error: Output file, $OptionsInfo{OutFileName}, already exists.\nUse \-o --overwrite\ option or specify a different name using \"-r --root\" option.\n";
|
|
|
293 }
|
|
|
294 }
|
|
|
295 }
|
|
|
296 }
|
|
|
297
|
|
|
298 # Setup properties to list...
|
|
|
299 sub SetupSpecifiedProperties {
|
|
|
300
|
|
|
301 $OptionsInfo{Properties} = defined $Options{properties} ? $Options{properties} : undef;
|
|
|
302
|
|
|
303 $OptionsInfo{PropertiesMode} = $Options{propertiesmode};
|
|
|
304 $OptionsInfo{PropertiesListing} = $Options{propertieslisting};
|
|
|
305
|
|
|
306 # Make sure appropriate properties/category names are specified...
|
|
|
307 @{$OptionsInfo{SpecifiedProperies}} = ();
|
|
|
308 if ($Options{properties} && ($Options{propertiesmode} =~ /^All$/i) ) {
|
|
|
309 warn "Warning: Ignoring values specifed by \"-p --properties\" option: Not valid for All value of \"--propertiesmode\" option...\n";
|
|
|
310 }
|
|
|
311 if ($Options{propertiesmode} =~ /^All$/i) {
|
|
|
312 if ($Options{propertieslisting} =~ /^Alphabetical$/i) {
|
|
|
313 push @{$OptionsInfo{SpecifiedProperies}}, NucleicAcids::GetNucleicAcidPropertiesNames('Alphabetical');
|
|
|
314 }
|
|
|
315 else {
|
|
|
316 push @{$OptionsInfo{SpecifiedProperies}}, NucleicAcids::GetNucleicAcidPropertiesNames();
|
|
|
317 }
|
|
|
318 }
|
|
|
319 else {
|
|
|
320 if ($Options{properties}) {
|
|
|
321 if ($Options{propertiesmode} =~ /^Categories$/i) {
|
|
|
322 # Check category name...
|
|
|
323 if ($Options{properties} !~ /^(Basic|BasicPlus)$/i) {
|
|
|
324 die "Error: The value specified, $Options{properties}, for option \"-p --properties\" in conjunction with \"Categories\" value for option \"--propertiesmode\" is not valid. Allowed values: Basic and BasicPlus\n";
|
|
|
325 }
|
|
|
326 # Set propertynames...
|
|
|
327 push @{$OptionsInfo{SpecifiedProperies}}, GetPropertyNamesFromCategories($Options{properties});
|
|
|
328 }
|
|
|
329 else {
|
|
|
330 # Check property names..
|
|
|
331 my($Name, $PropertyName, @Names);
|
|
|
332 @Names = split /\,/, $Options{properties};
|
|
|
333 NAME: for $Name (@Names) {
|
|
|
334 $PropertyName = RemoveLeadingAndTrailingWhiteSpaces($Name);
|
|
|
335 if (NucleicAcids::IsNucleicAcidProperty($PropertyName)) {
|
|
|
336 push @{$OptionsInfo{SpecifiedProperies}}, $PropertyName;
|
|
|
337 }
|
|
|
338 else {
|
|
|
339 warn "Warning: Ignoring value, $Name, specifed by \"-p --properties\" option: Unknown property name...\n";
|
|
|
340 }
|
|
|
341 }
|
|
|
342 if ($Options{propertieslisting} =~ /^Alphabetical$/i) {
|
|
|
343 # Code, OtherCodes and Name are always listed first...
|
|
|
344 my($CodePresent, $OtherCodesPresent, $NamePresent, @AlphabeticalProperties, %PropertiesMap);
|
|
|
345 %PropertiesMap = ();
|
|
|
346 @AlphabeticalProperties = ();
|
|
|
347 $CodePresent = 0; $OtherCodesPresent = 0; $NamePresent = 0;
|
|
|
348 NAME: for $Name (@{$OptionsInfo{SpecifiedProperies}}) {
|
|
|
349 if ($Name =~ /^Code$/i) {
|
|
|
350 $CodePresent = 1;
|
|
|
351 next NAME;
|
|
|
352 }
|
|
|
353 if ($Name =~ /^OtherCodes$/i) {
|
|
|
354 $OtherCodesPresent = 1;
|
|
|
355 next NAME;
|
|
|
356 }
|
|
|
357 if ($Name =~ /^Name$/i) {
|
|
|
358 $NamePresent = 1;
|
|
|
359 next NAME;
|
|
|
360 }
|
|
|
361 $PropertiesMap{$Name} = $Name;
|
|
|
362 }
|
|
|
363 # Setup the alphabetical list...
|
|
|
364 if ($CodePresent) {
|
|
|
365 push @AlphabeticalProperties, 'Code';
|
|
|
366 }
|
|
|
367 if ($OtherCodesPresent) {
|
|
|
368 push @AlphabeticalProperties, 'OtherCodesPresent';
|
|
|
369 }
|
|
|
370 if ($NamePresent) {
|
|
|
371 push @AlphabeticalProperties, 'Name';
|
|
|
372 }
|
|
|
373 for $Name (sort keys %PropertiesMap) {
|
|
|
374 push @AlphabeticalProperties, $Name;
|
|
|
375 }
|
|
|
376 @{$OptionsInfo{SpecifiedProperies}} = ();
|
|
|
377 push @{$OptionsInfo{SpecifiedProperies}}, @AlphabeticalProperties;
|
|
|
378 }
|
|
|
379 }
|
|
|
380 }
|
|
|
381 else {
|
|
|
382 # Set default value...
|
|
|
383 push @{$OptionsInfo{SpecifiedProperies}}, GetPropertyNamesFromCategories('Basic');
|
|
|
384 }
|
|
|
385 }
|
|
|
386 }
|
|
|
387
|
|
|
388 # Setup script usage and retrieve command line arguments specified using various options...
|
|
|
389 sub SetupScriptUsage {
|
|
|
390
|
|
|
391 # Retrieve all the options...
|
|
|
392 %Options = ();
|
|
|
393 $Options{mode} = "NucleicAcidID";
|
|
|
394 $Options{outdelim} = "comma";
|
|
|
395 $Options{output} = "STDOUT";
|
|
|
396 $Options{outputstyle} = "NucleicAcidBlock";
|
|
|
397 $Options{precision} = 4;
|
|
|
398 $Options{propertiesmode} = "Categories";
|
|
|
399 $Options{propertieslisting} = "ByGroup";
|
|
|
400 $Options{quote} = "yes";
|
|
|
401
|
|
|
402 if (!GetOptions(\%Options, "help|h", "mode|m=s", "outdelim=s", "output=s", "outputstyle=s", "overwrite|o", "precision=i", "properties|p=s", "propertieslisting=s", "propertiesmode=s", "quote|q=s", "root|r=s", "workingdir|w=s")) {
|
|
|
403 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
|
|
|
404 }
|
|
|
405 if ($Options{workingdir}) {
|
|
|
406 if (! -d $Options{workingdir}) {
|
|
|
407 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n";
|
|
|
408 }
|
|
|
409 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
|
|
|
410 }
|
|
|
411 if ($Options{mode} !~ /^(NucleicAcidID|NucleicAcidType)$/i) {
|
|
|
412 die "Error: The value specified, $Options{mode}, for option \"--mode\" is not valid. Allowed values: NucleicAcidID or NucleicAcidType\n";
|
|
|
413 }
|
|
|
414 if ($Options{outdelim} !~ /^(comma|semicolon|tab)$/i) {
|
|
|
415 die "Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n";
|
|
|
416 }
|
|
|
417 if ($Options{output} !~ /^(STDOUT|File)$/i) {
|
|
|
418 die "Error: The value specified, $Options{output}, for option \"--output\" is not valid. Allowed values: STDOUT or File\n";
|
|
|
419 }
|
|
|
420 if ($Options{outputstyle} !~ /^(NucleicAcidBlock|NucleicAcidRows)$/i) {
|
|
|
421 die "Error: The value specified, $Options{outputstyle}, for option \"--outputstyle\" is not valid. Allowed values: NucleicAcidBlock or NucleicAcidRows\n";
|
|
|
422 }
|
|
|
423 if (!IsPositiveInteger($Options{precision})) {
|
|
|
424 die "Error: The value specified, $Options{precision}, for option \"-p --precision\" is not valid. Allowed values: > 0 \n";
|
|
|
425 }
|
|
|
426 if ($Options{propertiesmode} !~ /^(Categories|Names|All)$/i) {
|
|
|
427 die "Error: The value specified, $Options{propertiesmode}, for option \"--propertiesmode\" is not valid. Allowed values: Categories, Names, or All\n";
|
|
|
428 }
|
|
|
429 if ($Options{propertieslisting} !~ /^(ByGroup|Alphabetical)$/i) {
|
|
|
430 die "Error: The value specified, $Options{propertieslisting}, for option \"--propertieslisting\" is not valid. Allowed values: ByGroup, or Alphabetical\n";
|
|
|
431 }
|
|
|
432 if ($Options{quote} !~ /^(yes|no)$/i) {
|
|
|
433 die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n";
|
|
|
434 }
|
|
|
435 }
|
|
|
436
|
|
|
437 __END__
|
|
|
438
|
|
|
439 =head1 NAME
|
|
|
440
|
|
|
441 InfoNucleicAcids.pl - List properties of nucleic acids
|
|
|
442
|
|
|
443 =head1 SYNOPSIS
|
|
|
444
|
|
|
445 InfoNucleicAcids.pl NucleicAcidIDs...
|
|
|
446
|
|
|
447 InfoNucleicAcids.pl [B<-h, --help>] [B<-m, --mode> NucleicAcidID | NucleicAcidType]
|
|
|
448 [B<--OutDelim> comma | tab | semicolon]
|
|
|
449 [B<--output> STDOUT | File] [B<--OutputStyle> NucleicAcidBlock | NucleicAcidRows]
|
|
|
450 [B<-o, --overwrite>] [B<--precision> number] [B<--PropertiesMode> Categories | Names | All]
|
|
|
451 [B<-p, --properties> CategoryName, [CategoryName,...] | PropertyName, [PropertyName,...]]
|
|
|
452 [B<--PropertiesListing> ByGroup | Alphabetical] [B<-q, --quote> yes | no] [B<-r, --root> rootname]
|
|
|
453 [B<-w, --WorkingDir> dirname] NucleicAcidIDs...
|
|
|
454
|
|
|
455 =head1 DESCRIPTION
|
|
|
456
|
|
|
457 List nucleic acid properties. Nucleic acids identification supports two types of IDs: code
|
|
|
458 or name. Nucleic acid properties data, in addition to basic information about nucleic acids - code,
|
|
|
459 name, type, chemical formula and molecular weight - include information about exact mass and
|
|
|
460 elemental composition.
|
|
|
461
|
|
|
462 =head1 PARAMETERS
|
|
|
463
|
|
|
464 =over 4
|
|
|
465
|
|
|
466 =item B<NucleicAcidIDs> I<Code [NucleicAcidName...] | NucleicAcidType [NucleicAcidType...]>
|
|
|
467
|
|
|
468 I<NucleicAcidIDs> is a space delimited list of values to identify nucleic acids.
|
|
|
469
|
|
|
470 For I<NucleicAcidID> mode, input value format is: I<Code [NucleicAcidName...]>. Default: I<A>.
|
|
|
471 Examples:
|
|
|
472
|
|
|
473 A
|
|
|
474 dG AMP
|
|
|
475 Cytidine T UDP dpppA "5'-dATP"
|
|
|
476
|
|
|
477 For I<NucleicAcidType> mode, input value format is: I<NucleicAcidType [NucleicAcidType...]>.
|
|
|
478 Default: I<A>. Possible values are: I<Nucleobase, Nucleoside, Deoxynucleoside, Nucleotide,
|
|
|
479 Deoxynucleotide>. Default: I<Nucleoside>.
|
|
|
480 Examples:
|
|
|
481
|
|
|
482 Deoxynucleoside
|
|
|
483 Nucleobase Nucleotide
|
|
|
484
|
|
|
485 =back
|
|
|
486
|
|
|
487 =head1 OPTIONS
|
|
|
488
|
|
|
489 =over 4
|
|
|
490
|
|
|
491 =item B<-h, --help>
|
|
|
492
|
|
|
493 Print this help message.
|
|
|
494
|
|
|
495 =item B<-m, --mode> I<NucleicAcidID | NucleicAcidType>
|
|
|
496
|
|
|
497 Specify nucleic acids for listing properties using one of these methods: nucleic acid
|
|
|
498 code and/or names or nucleic acid type.
|
|
|
499
|
|
|
500 Possible values: I<NucleicAcidID or NucleicAcidType>. Default: I<NucleicAcidID>
|
|
|
501
|
|
|
502 For I<NucleicAcidType>, command line parameters support these type: I<Nucleobase,
|
|
|
503 Nucleoside, Deoxynucleoside, Nucleotide, Deoxynucleotide>.
|
|
|
504
|
|
|
505 =item B<--OutDelim> I<comma | tab | semicolon>
|
|
|
506
|
|
|
507 Output text file delimiter. Possible values: I<comma, tab, or semicolon>
|
|
|
508 Default value: I<comma>.
|
|
|
509
|
|
|
510 =item B<--output> I<STDOUT | File>
|
|
|
511
|
|
|
512 List information at STDOUT or write it to a file. Possible values: I<STDOUT or File>. Default:
|
|
|
513 I<STDOUT>. B<-r, --root> option is used to generate output file name.
|
|
|
514
|
|
|
515 =item B<--OutputStyle> I<NucleicAcidBlock | NucleicAcidRows>
|
|
|
516
|
|
|
517 Specify how to list nucleic acid information: add a new line for each property and present it as a block
|
|
|
518 for each nucleic acid; or include all properties in one line and show it as a single line.
|
|
|
519
|
|
|
520 Possible values: I<NucleicAcidBlock | NucleicAcidRows>. Default: I<NucleicAcidBlock>
|
|
|
521
|
|
|
522 An example for I<NucleicAcidBlock> output style:
|
|
|
523
|
|
|
524 Code: Ado
|
|
|
525 OtherCodes: A
|
|
|
526 Name: Adenosine
|
|
|
527 Type: Nucleoside
|
|
|
528 MolecularFormula: C10H13O4N5
|
|
|
529 MolecularWeight: 267.2413
|
|
|
530 ... ...
|
|
|
531
|
|
|
532 An example for I<NucleicAcidRows> output style:
|
|
|
533
|
|
|
534 Code,OtherCodes,Name,Type,MolecularFormula,MolecularWeight
|
|
|
535
|
|
|
536 =item B<-o, --overwrite>
|
|
|
537
|
|
|
538 Overwrite existing files.
|
|
|
539
|
|
|
540 =item B<--precision> I<number>
|
|
|
541
|
|
|
542 Precision for listing numerical values. Default: up to I<4> decimal places.
|
|
|
543 Valid values: positive integers.
|
|
|
544
|
|
|
545 =item B<--PropertiesMode> I<Categories | Names | All>
|
|
|
546
|
|
|
547 Specify how property names are specified: use category names; explicit list of property names; or
|
|
|
548 use all available properties. Possible values: I<Categories, Names, or All>. Default: I<Categories>.
|
|
|
549
|
|
|
550 This option is used in conjunction with B<-p, --properties> option to specify properties of
|
|
|
551 interest.
|
|
|
552
|
|
|
553 =item B<-p, --properties> I<CategoryName,[CategoryName,...] | PropertyName,[PropertyName,...]>
|
|
|
554
|
|
|
555 This option is B<--propertiesmode> specific. In general, it's a list of comma separated category or
|
|
|
556 property names.
|
|
|
557
|
|
|
558 Specify which nucleic acid properties information to list for the nucleic acid IDs specified using
|
|
|
559 command line parameters: list basic information; list all available information; or specify a comma
|
|
|
560 separated list of nucleic acid property names.
|
|
|
561
|
|
|
562 Possible values: I<Basic | BasicPlus | PropertyName,[PropertyName,...]>.
|
|
|
563 Default: I<Basic>.
|
|
|
564
|
|
|
565 I<Basic> includes: I<Code, OtherCodes, Name, Type, MolecularFormula, MolecularWeight>
|
|
|
566
|
|
|
567 I<BasicPlus> includes: I<Code, OtherCodes, Name, Type, MolecularFormula, MolecularWeight, ExactMass,
|
|
|
568 ElementalComposition>
|
|
|
569
|
|
|
570 Here is a complete list of available properties: I<Code, OtherCodes, BasePair, Name, Type, MolecularFormula,
|
|
|
571 MolecularFormulaAtpH7.5, MolecularWeight, ExactMass, ElementalComposition>.
|
|
|
572
|
|
|
573 =item B<--PropertiesListing> I<ByGroup | Alphabetical>
|
|
|
574
|
|
|
575 Specify how to list properties for nucleic acids: group by category or an alphabetical by
|
|
|
576 property names. Possible values: I<ByGroup or Alphabetical>. Default: I<ByGroup>
|
|
|
577
|
|
|
578 =item B<-q, --quote> I<yes | no>
|
|
|
579
|
|
|
580 Put quotes around column values in output text file. Possible values: I<yes or
|
|
|
581 no>. Default value: I<yes>.
|
|
|
582
|
|
|
583 =item B<-r, --root> I<rootname>
|
|
|
584
|
|
|
585 New text file name is generated using the root: <Root>.<Ext>. File name is only
|
|
|
586 used during I<File> value of B<-o, --output> option.
|
|
|
587
|
|
|
588 Default file name: NucleicAcidInfo<mode>.<Ext>. The csv, and tsv
|
|
|
589 <Ext> values are used for comma/semicolon, and tab delimited text files respectively.
|
|
|
590
|
|
|
591 =item B<-w, --WorkingDir> I<dirname>
|
|
|
592
|
|
|
593 Location of working directory. Default: current directory.
|
|
|
594
|
|
|
595 =back
|
|
|
596
|
|
|
597 =head1 EXAMPLES
|
|
|
598
|
|
|
599 To list basic properties information for nucleoside A, type:
|
|
|
600
|
|
|
601 % InfoNucleicAcids.pl
|
|
|
602
|
|
|
603 To list all available properties information for nucleoside A, type:
|
|
|
604
|
|
|
605 % InfoNucleicAcids.pl --propertiesmode all A
|
|
|
606
|
|
|
607 To list all available information for all available nucleic acids, type:
|
|
|
608
|
|
|
609 % InfoNucleicAcids.pl --propertiesmode All All
|
|
|
610
|
|
|
611 To list basic properties information for all nucleobases, type:
|
|
|
612
|
|
|
613 % InfoNucleicAcids.pl -m NucleicAcidType Nucleoside
|
|
|
614
|
|
|
615 To list basic properties information for all nucleotides and deoxynulceotides, type:
|
|
|
616
|
|
|
617 % InfoNucleicAcids.pl -m NucleicAcidType Nucleotide Deoxynucleotide
|
|
|
618
|
|
|
619 To list basic properties information for variety of nucleic acids, type:
|
|
|
620
|
|
|
621 % InfoNucleicAcids.pl A dG AMP Cytidine T UDP "5'-dATP"
|
|
|
622
|
|
|
623 To list code and molecular weights for nucleosides A, G, C and T, type:
|
|
|
624
|
|
|
625 % InfoNucleicAcids.pl --PropertiesMode Names --properties
|
|
|
626 Code,MolecularWeight A G C T
|
|
|
627
|
|
|
628 To alphabetically list all the available properties for nucleotides dAMP, dGMP,
|
|
|
629 dCMP, and dTMP in rows instead of nucleic acid blocks with quotes around the values, type:
|
|
|
630
|
|
|
631 % InfoNucleicAcids.pl --PropertiesMode All --PropertiesListing
|
|
|
632 Alphabetical --OutputStyle NucleicAcidRows -q yes dAMP dGMP
|
|
|
633 dCMP dTMP
|
|
|
634
|
|
|
635 To alphabetically list all the available properties for all available nucleic acids to
|
|
|
636 a file names NucleicAcidsProperties.csv with quotes around the values, type
|
|
|
637
|
|
|
638 % InfoNucleicAcids.pl --PropertiesMode All --PropertiesListing
|
|
|
639 Alphabetical --output File --OutputStyle NucleicAcidRows -r
|
|
|
640 NucleicAcidsProperties -o -q Yes All
|
|
|
641
|
|
|
642 =head1 AUTHOR
|
|
|
643
|
|
|
644 Manish Sud <msud@san.rr.com>
|
|
|
645
|
|
|
646 =head1 SEE ALSO
|
|
|
647
|
|
|
648 InfoAminoAcids.pl, InfoPeriodicTableElements.pl
|
|
|
649
|
|
|
650 =head1 COPYRIGHT
|
|
|
651
|
|
|
652 Copyright (C) 2015 Manish Sud. All rights reserved.
|
|
|
653
|
|
|
654 This file is part of MayaChemTools.
|
|
|
655
|
|
|
656 MayaChemTools is free software; you can redistribute it and/or modify it under
|
|
|
657 the terms of the GNU Lesser General Public License as published by the Free
|
|
|
658 Software Foundation; either version 3 of the License, or (at your option)
|
|
|
659 any later version.
|
|
|
660
|
|
|
661 =cut
|
