|
0
|
1 #!/usr/bin/perl -w
|
|
|
2 #
|
|
|
3 # $RCSfile: InfoAminoAcids.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 "$FindBin::Bin/../lib";
|
|
|
31 use Getopt::Long;
|
|
|
32 use File::Basename;
|
|
|
33 use Text::ParseWords;
|
|
|
34 use Benchmark;
|
|
|
35 use FileUtil;
|
|
|
36 use TextUtil;
|
|
|
37 use AminoAcids;
|
|
|
38
|
|
|
39 my($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 ListAminoAcidProperties();
|
|
|
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 amino acid...
|
|
|
69 sub ListAminoAcidData {
|
|
|
70 my($DataLabelRef, $DataValueRef) = @_;
|
|
|
71 my($Index, $Line, $Value);
|
|
|
72
|
|
|
73 if ($OptionsInfo{AminoAcidRowsOutput}) {
|
|
|
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 "$Line\n";
|
|
|
107 }
|
|
|
108 }
|
|
|
109 }
|
|
|
110 }
|
|
|
111
|
|
|
112 # List data for an amino acid...
|
|
|
113 sub ListHeaderRowData {
|
|
|
114 my($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 amino acids...
|
|
|
130 sub ListAminoAcidProperties {
|
|
|
131 my($AminoAcidID, $AminoAcidDataRef, $PropertyName, $PropertyValue, @PropertyLabels, @PropertyValues);
|
|
|
132
|
|
|
133 print "Listing information for amino acid(s)...\n";
|
|
|
134
|
|
|
135 if ($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{AminoAcidRowsOutput}) {
|
|
|
147 ListHeaderRowData(\@PropertyLabels);
|
|
|
148 }
|
|
|
149
|
|
|
150 # Go over specified properties...
|
|
|
151 for $AminoAcidID (@{$OptionsInfo{SpecifiedAminoAcidIDs}}) {
|
|
|
152 $AminoAcidDataRef = AminoAcids::GetAminoAcidPropertiesData($AminoAcidID);
|
|
|
153
|
|
|
154 if (!$OptionsInfo{AminoAcidRowsOutput}) {
|
|
|
155 if ($OptionsInfo{FileOutput}) {
|
|
|
156 print OUTFILE "\nListing properties for amino acid $AminoAcidID...\n\n";
|
|
|
157 }
|
|
|
158 else {
|
|
|
159 print "\nListing properties for amino acid $AminoAcidID...\n\n";
|
|
|
160 }
|
|
|
161 }
|
|
|
162
|
|
|
163 # Collect data..
|
|
|
164 @PropertyValues = ();
|
|
|
165 for $PropertyName (@{$OptionsInfo{SpecifiedProperies}}) {
|
|
|
166 $PropertyValue = $AminoAcidDataRef->{$PropertyName};
|
|
|
167 if (IsFloat($PropertyValue)) {
|
|
|
168 $PropertyValue = sprintf("%.$OptionsInfo{Precision}f", $PropertyValue) + 0;
|
|
|
169 }
|
|
|
170 push @PropertyValues, $PropertyValue;
|
|
|
171 }
|
|
|
172 # List data...
|
|
|
173 ListAminoAcidData(\@PropertyLabels, \@PropertyValues);
|
|
|
174 }
|
|
|
175 if ($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$/i) {
|
|
|
188 @PropertyNames = ('ThreeLetterCode', 'OneLetterCode', 'AminoAcid', 'DNACodons', 'RNACodons', 'ChemicalFormula','MolecularWeight', 'LinearStructure', 'LinearStructureAtpH7.4');
|
|
|
189 } elsif ($CategoryName =~ /^BasicPlus$/i) {
|
|
|
190 @PropertyNames = ('ThreeLetterCode', 'OneLetterCode', 'AminoAcid', 'DNACodons', 'RNACodons', 'AcidicBasic', 'PolarNonpolar', 'Charged', 'Aromatic', 'HydrophobicHydophilic', 'IsoelectricPoint', 'pKCOOH', 'pKNH3+', 'ChemicalFormula', 'MolecularWeight', 'ExactMass', 'ChemicalFormulaMinusH2O', 'MolecularWeightMinusH2O(18.01524)', 'ExactMassMinusH2O(18.01056)','LinearStructure', 'LinearStructureAtpH7.4');
|
|
|
191 } elsif ($CategoryName =~ /^BasicAndHydrophobicity$/i) {
|
|
|
192 @PropertyNames = ('ThreeLetterCode', 'OneLetterCode', 'AminoAcid', 'DNACodons', 'RNACodons', 'ChemicalFormula', 'MolecularWeight', 'LinearStructure', 'LinearStructureAtpH7.4', 'HydrophobicityEisenbergAndOthers', 'HydrophobicityHoppAndWoods', 'HydrophobicityJanin', 'HydrophobicityKyteAndDoolittle', 'HydrophobicityRoseAndOthers', 'HydrophobicityWolfendenAndOthers');
|
|
|
193 } elsif ($CategoryName =~ /^BasicAndHydrophobicityPlus$/i) {
|
|
|
194 @PropertyNames = ('ThreeLetterCode', 'OneLetterCode', 'AminoAcid', 'DNACodons', 'RNACodons', 'ChemicalFormula', 'MolecularWeight', 'LinearStructure', 'LinearStructureAtpH7.4', 'HydrophobicityAbrahamAndLeo', 'HydrophobicityBlack', 'HydrophobicityBullAndBreese', 'HydrophobicityChothia', 'HydrophobicityEisenbergAndOthers', 'HydrophobicityFauchereAndOthers', 'HydrophobicityGuy', 'HydrophobicityHPLCAtpH3.4Cowan', 'HydrophobicityHPLCAtpH7.5Cowan', 'HydrophobicityHPLCParkerAndOthers', 'HydrophobicityHPLCWilsonAndOthers', 'HydrophobicityHoppAndWoods', 'HydrophobicityJanin', 'HydrophobicityKyteAndDoolittle', 'HydrophobicityManavalanAndOthers', 'HydrophobicityMiyazawaAndOthers', 'HydrophobicityOMHSweetAndOthers', 'HydrophobicityRaoAndArgos', 'HydrophobicityRfMobility', 'HydrophobicityRoseAndOthers', 'HydrophobicityRoseman', 'HydrophobicityWellingAndOthers', 'HydrophobicityWolfendenAndOthers');
|
|
|
195 }
|
|
|
196
|
|
|
197 return @PropertyNames;
|
|
|
198 }
|
|
|
199
|
|
|
200 # Process option values...
|
|
|
201 sub ProcessOptions {
|
|
|
202 %OptionsInfo = ();
|
|
|
203
|
|
|
204 $OptionsInfo{OutDelim} = ($Options{outdelim} =~ /^tab$/i ) ? "\t" : (($Options{outdelim} =~ /^semicolon$/i) ? "\;" : "\,");
|
|
|
205 $OptionsInfo{OutQuote} = ($Options{quote} =~ /^yes$/i) ? 1 : 0;
|
|
|
206
|
|
|
207 $OptionsInfo{Overwrite} = defined $Options{overwrite} ? $Options{overwrite} : undef;
|
|
|
208 $OptionsInfo{OutFileRoot} = defined $Options{root} ? $Options{root} : undef;
|
|
|
209
|
|
|
210 $OptionsInfo{OutputStyle} = $Options{outputstyle};
|
|
|
211
|
|
|
212 $OptionsInfo{AminoAcidRowsOutput} = ($Options{outputstyle} =~ /^AminoAcidRows$/i) ? 1 : 0;
|
|
|
213 $OptionsInfo{FileOutput} = ($Options{output} =~ /^File$/i) ? 1 : 0;
|
|
|
214
|
|
|
215 $OptionsInfo{Precision} = $Options{precision};
|
|
|
216
|
|
|
217 my($AminoAcidID, @AminoAcidIDs);
|
|
|
218
|
|
|
219 @{$OptionsInfo{SpecifiedAminoAcidIDs}} = ();
|
|
|
220
|
|
|
221 # Set up Amino Acids IDs except for All mode...
|
|
|
222 @AminoAcidIDs = ();
|
|
|
223
|
|
|
224 if (@ARGV >= 1) {
|
|
|
225 push @AminoAcidIDs, @ARGV;
|
|
|
226 }
|
|
|
227 else {
|
|
|
228 # Setup mode specified default values...
|
|
|
229 push @AminoAcidIDs, 'Ala';
|
|
|
230 }
|
|
|
231
|
|
|
232 # Generate list of amino acids...
|
|
|
233 if (@ARGV == 1 && $ARGV[0] =~ /^All$/i) {
|
|
|
234 push @{$OptionsInfo{SpecifiedAminoAcidIDs}}, AminoAcids::GetAminoAcids();
|
|
|
235 }
|
|
|
236 else {
|
|
|
237 ID: for $AminoAcidID (@AminoAcidIDs) {
|
|
|
238 if (AminoAcids::IsAminoAcid($AminoAcidID)) {
|
|
|
239 push @{$OptionsInfo{SpecifiedAminoAcidIDs}}, $AminoAcidID;
|
|
|
240 }
|
|
|
241 else {
|
|
|
242 warn "Ignoring amino acid ID, $AminoAcidID, specified using command line parameter option: Unknown amino acid ID...\n";
|
|
|
243 next ID;
|
|
|
244 }
|
|
|
245 }
|
|
|
246 }
|
|
|
247 SetupSpecifiedProperties();
|
|
|
248
|
|
|
249 # Setup output file name...
|
|
|
250 $OptionsInfo{OutFileName} = '';
|
|
|
251 if ($OptionsInfo{FileOutput}) {
|
|
|
252 my($OutFileRoot, $OutFileExt);
|
|
|
253
|
|
|
254 $OutFileRoot = '';
|
|
|
255 $OutFileExt = "csv";
|
|
|
256 if ($Options{outdelim} =~ /^tab$/i) {
|
|
|
257 $OutFileExt = "tsv";
|
|
|
258 }
|
|
|
259 if ($Options{root}) {
|
|
|
260 my ($RootFileDir, $RootFileName, $RootFileExt) = ParseFileName($Options{root});
|
|
|
261 if ($RootFileName && $RootFileExt) {
|
|
|
262 $OutFileRoot = $RootFileName;
|
|
|
263 }
|
|
|
264 else {
|
|
|
265 $OutFileRoot = $Options{root};
|
|
|
266 }
|
|
|
267 }
|
|
|
268 else {
|
|
|
269 $OutFileRoot = 'AminoAcidsInfo';
|
|
|
270 }
|
|
|
271 $OptionsInfo{OutFileName} = $OutFileRoot . '.' . $OutFileExt;
|
|
|
272 if (!$Options{overwrite}) {
|
|
|
273 if (-e $OptionsInfo{OutFileName}) {
|
|
|
274 die "Error: Output file, $OptionsInfo{OutFileName}, already exists.\nUse \-o --overwrite\ option or specify a different name using \"-r --root\" option.\n";
|
|
|
275 }
|
|
|
276 }
|
|
|
277 }
|
|
|
278 }
|
|
|
279
|
|
|
280 # Setup properties to list...
|
|
|
281 sub SetupSpecifiedProperties {
|
|
|
282
|
|
|
283 $OptionsInfo{Properties} = defined $Options{properties} ? $Options{properties} : undef;
|
|
|
284
|
|
|
285 $OptionsInfo{PropertiesMode} = $Options{propertiesmode};
|
|
|
286 $OptionsInfo{PropertiesListing} = $Options{propertieslisting};
|
|
|
287
|
|
|
288 # Make sure appropriate properties/category names are specified...
|
|
|
289 @{$OptionsInfo{SpecifiedProperies}} = ();
|
|
|
290 if ($Options{properties} && ($Options{propertiesmode} =~ /^All$/i) ) {
|
|
|
291 warn "Warning: Ignoring values specifed by \"-p --properties\" option: Not valid for All value of \"--propertiesmode\" option...\n";
|
|
|
292 }
|
|
|
293 if ($Options{propertiesmode} =~ /^All$/i) {
|
|
|
294 if ($Options{propertieslisting} =~ /^Alphabetical$/i) {
|
|
|
295 push @{$OptionsInfo{SpecifiedProperies}}, AminoAcids::GetAminoAcidPropertiesNames('Alphabetical');
|
|
|
296 }
|
|
|
297 else {
|
|
|
298 push @{$OptionsInfo{SpecifiedProperies}}, AminoAcids::GetAminoAcidPropertiesNames();
|
|
|
299 }
|
|
|
300 }
|
|
|
301 else {
|
|
|
302 if ($Options{properties}) {
|
|
|
303 if ($Options{propertiesmode} =~ /^Categories$/i) {
|
|
|
304 # Check category name...
|
|
|
305 if ($Options{properties} !~ /^(Basic|BasicPlus|BasicAndHydrophobicity|BasicAndHydrophobicityPlus)$/i) {
|
|
|
306 die "Error: The value specified, $Options{properties}, for option \"-p --properties\" in conjunction with \"Categories\" value for option \"--propertiesmode\" is not valid. Allowed values: Basic, BasicPlus, BasicAndHydrophobicity, and BasicAndHydrophobicityPlus\n";
|
|
|
307 }
|
|
|
308 # Set propertynames...
|
|
|
309 push @{$OptionsInfo{SpecifiedProperies}}, GetPropertyNamesFromCategories($Options{properties});
|
|
|
310 }
|
|
|
311 else {
|
|
|
312 # Check property names..
|
|
|
313 my($Name, $PropertyName, @Names);
|
|
|
314 @Names = split /\,/, $Options{properties};
|
|
|
315 NAME: for $Name (@Names) {
|
|
|
316 $PropertyName = RemoveLeadingAndTrailingWhiteSpaces($Name);
|
|
|
317 if (AminoAcids::IsAminoAcidProperty($PropertyName)) {
|
|
|
318 push @{$OptionsInfo{SpecifiedProperies}}, $PropertyName;
|
|
|
319 }
|
|
|
320 else {
|
|
|
321 warn "Warning: Ignoring value, $Name, specifed by \"-p --properties\" option: Unknown property name...\n";
|
|
|
322 }
|
|
|
323 }
|
|
|
324 if ($Options{propertieslisting} =~ /^Alphabetical$/i) {
|
|
|
325 # ThreeLetterCode, OneLetterCode and AminoAcid are always listed first...
|
|
|
326 # NaturalIsotopeData in the end...
|
|
|
327 my($OneLetterCodePresent, $ThreeLetterCodePresent, $AminoAcidPresent, @AlphabeticalProperties, %PropertiesMap);
|
|
|
328 %PropertiesMap = ();
|
|
|
329 @AlphabeticalProperties = ();
|
|
|
330 $OneLetterCodePresent = 0; $ThreeLetterCodePresent = 0; $AminoAcidPresent = 0;
|
|
|
331 NAME: for $Name (@{$OptionsInfo{SpecifiedProperies}}) {
|
|
|
332 if ($Name =~ /^OneLetterCode$/i) {
|
|
|
333 $OneLetterCodePresent = 1;
|
|
|
334 next NAME;
|
|
|
335 }
|
|
|
336 if ($Name =~ /^ThreeLetterCode$/i) {
|
|
|
337 $ThreeLetterCodePresent = 1;
|
|
|
338 next NAME;
|
|
|
339 }
|
|
|
340 if ($Name =~ /^AminoAcid$/i) {
|
|
|
341 $AminoAcidPresent = 1;
|
|
|
342 next NAME;
|
|
|
343 }
|
|
|
344 $PropertiesMap{$Name} = $Name;
|
|
|
345 }
|
|
|
346 # Setup the alphabetical list...
|
|
|
347 if ($ThreeLetterCodePresent) {
|
|
|
348 push @AlphabeticalProperties, 'ThreeLetterCode';
|
|
|
349 }
|
|
|
350 if ($OneLetterCodePresent) {
|
|
|
351 push @AlphabeticalProperties, 'OneLetterCode';
|
|
|
352 }
|
|
|
353 if ($AminoAcidPresent) {
|
|
|
354 push @AlphabeticalProperties, 'AminoAcid';
|
|
|
355 }
|
|
|
356 for $Name (sort keys %PropertiesMap) {
|
|
|
357 push @AlphabeticalProperties, $Name;
|
|
|
358 }
|
|
|
359 @{$OptionsInfo{SpecifiedProperies}} = ();
|
|
|
360 push @{$OptionsInfo{SpecifiedProperies}}, @AlphabeticalProperties;
|
|
|
361 }
|
|
|
362 }
|
|
|
363 }
|
|
|
364 else {
|
|
|
365 # Set default value...
|
|
|
366 push @{$OptionsInfo{SpecifiedProperies}}, GetPropertyNamesFromCategories('Basic');
|
|
|
367 }
|
|
|
368 }
|
|
|
369 }
|
|
|
370
|
|
|
371 # Setup script usage and retrieve command line arguments specified using various options...
|
|
|
372 sub SetupScriptUsage {
|
|
|
373
|
|
|
374 # Retrieve all the options...
|
|
|
375 %Options = ();
|
|
|
376 $Options{outdelim} = "comma";
|
|
|
377 $Options{output} = "STDOUT";
|
|
|
378 $Options{outputstyle} = "AminoAcidBlock";
|
|
|
379 $Options{precision} = 4;
|
|
|
380 $Options{propertiesmode} = "Categories";
|
|
|
381 $Options{propertieslisting} = "ByGroup";
|
|
|
382 $Options{quote} = "yes";
|
|
|
383
|
|
|
384 if (!GetOptions(\%Options, "help|h", "outdelim=s", "output=s", "outputstyle=s", "overwrite|o", "precision=i", "properties|p=s", "propertieslisting=s", "propertiesmode=s", "quote|q=s", "root|r=s", "workingdir|w=s")) {
|
|
|
385 die "\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n";
|
|
|
386 }
|
|
|
387 if ($Options{workingdir}) {
|
|
|
388 if (! -d $Options{workingdir}) {
|
|
|
389 die "Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n";
|
|
|
390 }
|
|
|
391 chdir $Options{workingdir} or die "Error: Couldn't chdir $Options{workingdir}: $! \n";
|
|
|
392 }
|
|
|
393 if ($Options{outdelim} !~ /^(comma|semicolon|tab)$/i) {
|
|
|
394 die "Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n";
|
|
|
395 }
|
|
|
396 if ($Options{output} !~ /^(STDOUT|File)$/i) {
|
|
|
397 die "Error: The value specified, $Options{output}, for option \"--output\" is not valid. Allowed values: STDOUT or File\n";
|
|
|
398 }
|
|
|
399 if ($Options{outputstyle} !~ /^(AminoAcidBlock|AminoAcidRows)$/i) {
|
|
|
400 die "Error: The value specified, $Options{outputstyle}, for option \"--outputstyle\" is not valid. Allowed values: AminoAcidBlock or AminoAcidRows\n";
|
|
|
401 }
|
|
|
402 if (!IsPositiveInteger($Options{precision})) {
|
|
|
403 die "Error: The value specified, $Options{precision}, for option \"-p --precision\" is not valid. Allowed values: > 0 \n";
|
|
|
404 }
|
|
|
405 if ($Options{propertiesmode} !~ /^(Categories|Names|All)$/i) {
|
|
|
406 die "Error: The value specified, $Options{propertiesmode}, for option \"--propertiesmode\" is not valid. Allowed values: Categories, Names, or All\n";
|
|
|
407 }
|
|
|
408 if ($Options{propertieslisting} !~ /^(ByGroup|Alphabetical)$/i) {
|
|
|
409 die "Error: The value specified, $Options{propertieslisting}, for option \"--propertieslisting\" is not valid. Allowed values: ByGroup, or Alphabetical\n";
|
|
|
410 }
|
|
|
411 if ($Options{quote} !~ /^(yes|no)$/i) {
|
|
|
412 die "Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n";
|
|
|
413 }
|
|
|
414 }
|
|
|
415
|
|
|
416 __END__
|
|
|
417
|
|
|
418 =head1 NAME
|
|
|
419
|
|
|
420 InfoAminoAcids.pl - List properties of amino acids
|
|
|
421
|
|
|
422 =head1 SYNOPSIS
|
|
|
423
|
|
|
424 InfoAminoAcids.pl AminoAcidIDs...
|
|
|
425
|
|
|
426 InfoAminoAcids.pl [B<-h, --help>] [B<--outdelim> comma | tab | semicolon]
|
|
|
427 [B<--output> STDOUT | File] [B<--outputstyle> AminoAcidBlock | AminoAcidRows]
|
|
|
428 [B<-o, --overwrite>] [B<--precision> number] [B<--propertiesmode> Categories | Names | All]
|
|
|
429 [B<-p, --properties> CategoryName,[CategoryName,...] | PropertyName,[PropertyName,...]]
|
|
|
430 [B<--propertieslinting> ByGroup | Alphabetical] [B<-q, --quote> yes | no] [B<-r, --root> rootname]
|
|
|
431 [B<-w, --workingdir> dirname] AminoAcidIDs...
|
|
|
432
|
|
|
433 =head1 DESCRIPTION
|
|
|
434
|
|
|
435 List amino acid properties. Amino acids identification supports these three types of IDs: one letter
|
|
|
436 code, three letter code or name. Amino acid properties data, in addition to basic information about
|
|
|
437 amino acids - one and three letter codes, name, DNA and RNA codons, molecular weight - include
|
|
|
438 variety of other properties: polarity, acidity, hydrophobicity, and so on.
|
|
|
439
|
|
|
440 =head1 PARAMETERS
|
|
|
441
|
|
|
442 =over 4
|
|
|
443
|
|
|
444 =item B<AminoAcidIDs> I<ThreeLetterCode [OneLetterCode AminoAcidName...]>
|
|
|
445
|
|
|
446 I<AminoAcidIDs> is a space delimited list of values to identify amino acids.
|
|
|
447
|
|
|
448 Input value format is: I<ThreeLetterCode [OneLetterCode AminoAcidName...]>. Default: I<Ala>.
|
|
|
449 Examples:
|
|
|
450
|
|
|
451 Ala
|
|
|
452 Glu A
|
|
|
453 Alanine Glu Y "Aspartic acid"
|
|
|
454
|
|
|
455 =back
|
|
|
456
|
|
|
457 =head1 OPTIONS
|
|
|
458
|
|
|
459 =over 4
|
|
|
460
|
|
|
461 =item B<-h, --help>
|
|
|
462
|
|
|
463 Print this help message.
|
|
|
464
|
|
|
465 =item B<--outdelim> I<comma | tab | semicolon>
|
|
|
466
|
|
|
467 Output text file delimiter. Possible values: I<comma, tab, or semicolon>
|
|
|
468 Default value: I<comma>.
|
|
|
469
|
|
|
470 =item B<--output> I<STDOUT | File>
|
|
|
471
|
|
|
472 List information at STDOUT or write it to a file. Possible values: I<STDOUT or File>. Default:
|
|
|
473 I<STDOUT>. B<-r, --root> option is used to generate output file name.
|
|
|
474
|
|
|
475 =item B<--outputstyle> I<AminoAcidBlock | AminoAcidRows>
|
|
|
476
|
|
|
477 Specify how to list amino acid information: add a new line for each property and present it as a block
|
|
|
478 for each amino acid; or include all properties in one line and show it as a single line.
|
|
|
479
|
|
|
480 Possible values: I<AminoAcidBlock | AminoAcidRows>. Default: I<AminoAcidBlock>
|
|
|
481
|
|
|
482 An example for I<AminoAcidBlock> output style:
|
|
|
483
|
|
|
484 ThreeLetterCode: Ala
|
|
|
485 OneLetterCode: A
|
|
|
486 AminoAcid: Alanine
|
|
|
487 MolecularWeight: 89.0941
|
|
|
488 ... ...
|
|
|
489 ... ...
|
|
|
490 ... ...
|
|
|
491
|
|
|
492 ThreeLetterCode: Glu
|
|
|
493 OneLetterCode: E
|
|
|
494 AminoAcid: Glutamic acid
|
|
|
495 MolecularWeight: 147.1308
|
|
|
496 ... ...
|
|
|
497 ... ...
|
|
|
498 ... ...
|
|
|
499
|
|
|
500 An example for I<AminoAcidRows> output style:
|
|
|
501
|
|
|
502 ThreeLetterCode,OneLetterCode,AminoAcid,MolecularWeight
|
|
|
503 Ala,A,Alanine,89.0941..
|
|
|
504 Glu,E,Glutamic acid,147.1308..
|
|
|
505
|
|
|
506 =item B<-o, --overwrite>
|
|
|
507
|
|
|
508 Overwrite existing files.
|
|
|
509
|
|
|
510 =item B<--precision> I<number>
|
|
|
511
|
|
|
512 Precision for listing numerical values. Default: up to I<4> decimal places.
|
|
|
513 Valid values: positive integers.
|
|
|
514
|
|
|
515 =item B<--propertiesmode> I<Categories | Names | All>
|
|
|
516
|
|
|
517 Specify how property names are specified: use category names; explicit list of property names; or
|
|
|
518 use all available properties. Possible values: I<Categories, Names, or All>. Default: I<Categories>.
|
|
|
519
|
|
|
520 This option is used in conjunction with B<-p, --properties> option to specify properties of
|
|
|
521 interest.
|
|
|
522
|
|
|
523 =item B<-p, --properties> I<CategoryName,[CategoryName,...] | PropertyName,[PropertyName,...]>
|
|
|
524
|
|
|
525 This option is B<--propertiesmode> specific. In general, it's a list of comma separated category or
|
|
|
526 property names.
|
|
|
527
|
|
|
528 Specify which amino acid properties information to list for the amino acid IDs specified using command:
|
|
|
529 line parameters: list basic and/or hydrophobicity information; list all available information; or specify a comma
|
|
|
530 separated list of amino acid property names.
|
|
|
531
|
|
|
532 Possible values: I<Basic | BasicPlus | BasicAndHydrophobicity | BasicAndHydrophobicityPlus | PropertyName,[PropertyName,...]>.
|
|
|
533 Default: I<Basic>.
|
|
|
534
|
|
|
535 I<Basic> includes: I<ThreeLetterCode, OneLetterCode, AminoAcid, DNACodons, RNACodons, ChemicalFormula, MolecularWeight, LinearStructure, LinearStructureAtpH7.4>
|
|
|
536
|
|
|
537 I<BasicPlus> includes: I<ThreeLetterCode, OneLetterCode, AminoAcid, DNACodons, RNACodons, AcidicBasic, PolarNonpolar, Charged, Aromatic, HydrophobicHydophilic, IsoelectricPoint, pKCOOH, pKNH3+, ChemicalFormula, MolecularWeight, ExactMass, ChemicalFormulaMinusH2O, MolecularWeightMinusH2O(18.01524), ExactMassMinusH2O(18.01056), LinearStructure, LinearStructureAtpH7.4>
|
|
|
538
|
|
|
539 I<BasicAndHydrophobicity> includes: I<ThreeLetterCode, OneLetterCode, AminoAcid, DNACodons, RNACodons, ChemicalFormula, MolecularWeight, LinearStructure, LinearStructureAtpH7.4, HydrophobicityEisenbergAndOthers, HydrophobicityHoppAndWoods, HydrophobicityJanin, HydrophobicityKyteAndDoolittle, HydrophobicityRoseAndOthers, HydrophobicityWolfendenAndOthers>
|
|
|
540
|
|
|
541 I<BasicAndHydrophobicityPlus> includes: I<(ThreeLetterCode, OneLetterCode, AminoAcid, DNACodons, RNACodons, ChemicalFormula, MolecularWeight, LinearStructure, LinearStructureAtpH7.4, HydrophobicityAbrahamAndLeo, HydrophobicityBlack, HydrophobicityBullAndBreese, HydrophobicityChothia, HydrophobicityEisenbergAndOthers, HydrophobicityFauchereAndOthers, HydrophobicityGuy, HydrophobicityHPLCAtpH3.4Cowan, HydrophobicityHPLCAtpH7.5Cowan, HydrophobicityHPLCParkerAndOthers, HydrophobicityHPLCWilsonAndOthers, HydrophobicityHoppAndWoods, HydrophobicityJanin, HydrophobicityKyteAndDoolittle, HydrophobicityManavalanAndOthers, HydrophobicityMiyazawaAndOthers, HydrophobicityOMHSweetAndOthers, HydrophobicityRaoAndArgos, HydrophobicityRfMobility, HydrophobicityRoseAndOthers, HydrophobicityRoseman, HydrophobicityWellingAndOthers, HydrophobicityWolfendenAndOthers>
|
|
|
542
|
|
|
543 Here is a complete list of available properties: ThreeLetterCode, OneLetterCode, AminoAcid, DNACodons, RNACodons, AcidicBasic, PolarNonpolar, Charged, Aromatic, HydrophobicHydophilic, IsoelectricPoint, pKCOOH, pKNH3+, ChemicalFormula, MolecularWeight, ExactMass, ChemicalFormulaMinusH2O, MolecularWeightMinusH2O(18.01524), ExactMassMinusH2O(18.01056), vanderWaalsVolume, %AccessibleResidues, %BuriedResidues, AlphaHelixChouAndFasman, AlphaHelixDeleageAndRoux, AlphaHelixLevitt, AminoAcidsComposition, AminoAcidsCompositionInSwissProt, AntiparallelBetaStrand, AverageAreaBuried, AverageFlexibility, BetaSheetChouAndFasman, BetaSheetDeleageAndRoux, BetaSheetLevitt, BetaTurnChouAndFasman, BetaTurnDeleageAndRoux, BetaTurnLevitt, Bulkiness, CoilDeleageAndRoux, HPLCHFBARetention, HPLCRetentionAtpH2.1, HPLCRetentionAtpH7.4, HPLCTFARetention, HydrophobicityAbrahamAndLeo, HydrophobicityBlack, HydrophobicityBullAndBreese, HydrophobicityChothia, HydrophobicityEisenbergAndOthers, HydrophobicityFauchereAndOthers, HydrophobicityGuy, HydrophobicityHPLCAtpH3.4Cowan, HydrophobicityHPLCAtpH7.5Cowan, HydrophobicityHPLCParkerAndOthers, HydrophobicityHPLCWilsonAndOthers, HydrophobicityHoppAndWoods, HydrophobicityJanin, HydrophobicityKyteAndDoolittle, HydrophobicityManavalanAndOthers, HydrophobicityMiyazawaAndOthers, HydrophobicityOMHSweetAndOthers, HydrophobicityRaoAndArgos, HydrophobicityRfMobility, HydrophobicityRoseAndOthers, HydrophobicityRoseman, HydrophobicityWellingAndOthers, HydrophobicityWolfendenAndOthers, ParallelBetaStrand, PolarityGrantham, PolarityZimmerman, RatioHeteroEndToSide, RecognitionFactors, Refractivity, RelativeMutability, TotalBetaStrand, LinearStructure, LinearStructureAtpH7.4
|
|
|
544
|
|
|
545 =item B<--propertieslisting> I<ByGroup | Alphabetical>
|
|
|
546
|
|
|
547 Specify how to list properties for amino acids: group by category or an alphabetical by
|
|
|
548 property names. Possible values: I<ByGroup or Alphabetical>. Default: I<ByGroup>.
|
|
|
549
|
|
|
550 =item B<-q, --quote> I<yes | no>
|
|
|
551
|
|
|
552 Put quotes around column values in output text file. Possible values: I<yes or
|
|
|
553 no>. Default value: I<yes>.
|
|
|
554
|
|
|
555 =item B<-r, --root> I<rootname>
|
|
|
556
|
|
|
557 New text file name is generated using the root: <Root>.<Ext>. File name is only
|
|
|
558 used during I<File> value of B<-o, --output> option.
|
|
|
559
|
|
|
560 Default file name: AminoAcidInfo<mode>.<Ext>. The csv, and tsv
|
|
|
561 <Ext> values are used for comma/semicolon, and tab delimited text files respectively.
|
|
|
562
|
|
|
563 =item B<-w, --workingdir> I<dirname>
|
|
|
564
|
|
|
565 Location of working directory. Default: current directory.
|
|
|
566
|
|
|
567 =back
|
|
|
568
|
|
|
569 =head1 EXAMPLES
|
|
|
570
|
|
|
571 To list basic properties information for amino acid Ala, type:
|
|
|
572
|
|
|
573 % InfoAminoAcids.pl
|
|
|
574
|
|
|
575 To list all available properties information for amino acid Ala, type:
|
|
|
576
|
|
|
577 % InfoAminoAcids.pl --propertiesmode all
|
|
|
578
|
|
|
579 To list basic properties information for amino acids Ala, Arg, and Asp type:
|
|
|
580
|
|
|
581 % InfoAminoAcids.pl Ala Arg Asp
|
|
|
582 % InfoAminoAcids.pl A Arg "Aspartic acid"
|
|
|
583
|
|
|
584 To list all available properties information for amino acids Ala, Arg, and Asp type:
|
|
|
585
|
|
|
586 % InfoAminoAcids.pl --propertiesmode all Ala Arg Asp
|
|
|
587
|
|
|
588 To list basic and hydrophobicty properties information for amino acids Ala, Arg, and Asp type:
|
|
|
589
|
|
|
590 % InfoAminoAcids.pl --propertiesmode Categories
|
|
|
591 --properties BasicAndHydrophobicity Ala Arg Asp
|
|
|
592
|
|
|
593 To list OneLetterCode, ThreeLetterCode, DNACodons, and MolecularWeight for amino
|
|
|
594 acids Ala, Arg, and Asp type:
|
|
|
595
|
|
|
596 % InfoAminoAcids.pl --propertiesmode Names
|
|
|
597 --properties OneLetterCode,ThreeLetterCode,DNACodons,MolecularWeight
|
|
|
598 Ala Arg Asp
|
|
|
599
|
|
|
600 To alphabetically list basic and hydrophobicty properties information for amino acids Ala, Arg, and Asp
|
|
|
601 in rows insetad of amino acid blocks with quotes around the values, type:
|
|
|
602
|
|
|
603 % InfoAminoAcids.pl --propertiesmode Categories
|
|
|
604 --properties BasicAndHydrophobicity --propertieslisting alphabetical
|
|
|
605 --outdelim comma --outputstyle AminoAcidRows --quote yes Ala Arg Asp
|
|
|
606
|
|
|
607 To alphabetically list basic and hydrophobicty properties information for amino acids Ala, Arg, and Asp
|
|
|
608 in rows insetad of amino acid blocks with quotes around the values and write them into a file
|
|
|
609 AminoAcidProperties.csv, type:
|
|
|
610
|
|
|
611 % InfoAminoAcids.pl --propertiesmode Categories
|
|
|
612 --properties BasicAndHydrophobicity --propertieslisting alphabetical
|
|
|
613 --outdelim comma --outputstyle AminoAcidRows --quote yes
|
|
|
614 --output File -r AminoAcidProperties -o Ala Arg Asp
|
|
|
615
|
|
|
616 =head1 AUTHOR
|
|
|
617
|
|
|
618 Manish Sud <msud@san.rr.com>
|
|
|
619
|
|
|
620 =head1 SEE ALSO
|
|
|
621
|
|
|
622 InfoNucleicAcids.pl InfoPeriodicTableElements.pl
|
|
|
623
|
|
|
624 =head1 COPYRIGHT
|
|
|
625
|
|
|
626 Copyright (C) 2015 Manish Sud. All rights reserved.
|
|
|
627
|
|
|
628 This file is part of MayaChemTools.
|
|
|
629
|
|
|
630 MayaChemTools is free software; you can redistribute it and/or modify it under
|
|
|
631 the terms of the GNU Lesser General Public License as published by the Free
|
|
|
632 Software Foundation; either version 3 of the License, or (at your option)
|
|
|
633 any later version.
|
|
|
634
|
|
|
635 =cut
|
