|
0
|
1 package MolecularDescriptors::MolecularVolumeDescriptors;
|
|
|
2 #
|
|
|
3 # $RCSfile: MolecularVolumeDescriptors.pm,v $
|
|
|
4 # $Date: 2015/02/28 20:49:20 $
|
|
|
5 # $Revision: 1.16 $
|
|
|
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 Carp;
|
|
|
31 use Exporter;
|
|
|
32 use Scalar::Util ();
|
|
|
33 use TextUtil ();
|
|
|
34 use MathUtil ();
|
|
|
35 use Atom;
|
|
|
36 use Molecule;
|
|
|
37 use AtomTypes::AtomTypes;
|
|
|
38 use MolecularDescriptors::MolecularDescriptors;
|
|
|
39
|
|
|
40 use vars qw(@ISA @EXPORT @EXPORT_OK %EXPORT_TAGS);
|
|
|
41
|
|
|
42 @ISA = qw(MolecularDescriptors::MolecularDescriptors Exporter);
|
|
|
43 @EXPORT = qw();
|
|
|
44 @EXPORT_OK = qw(GetDescriptorNames GetVDWAtomRadiiAndVolumesData);
|
|
|
45
|
|
|
46 %EXPORT_TAGS = (all => [@EXPORT, @EXPORT_OK]);
|
|
|
47
|
|
|
48 # Setup class variables...
|
|
|
49 my($ClassName, @DescriptorNames, %VDWAtomRadiiAndVolumesDataMap);
|
|
|
50 _InitializeClass();
|
|
|
51
|
|
|
52 # Overload Perl functions...
|
|
|
53 use overload '""' => 'StringifyMolecularVolumeDescriptors';
|
|
|
54
|
|
|
55 # Class constructor...
|
|
|
56 sub new {
|
|
|
57 my($Class, %NamesAndValues) = @_;
|
|
|
58
|
|
|
59 # Initialize object...
|
|
|
60 my $This = $Class->SUPER::new();
|
|
|
61 bless $This, ref($Class) || $Class;
|
|
|
62 $This->_InitializeMolecularVolumeDescriptors();
|
|
|
63
|
|
|
64 $This->_InitializeMolecularVolumeDescriptorsProperties(%NamesAndValues);
|
|
|
65
|
|
|
66 return $This;
|
|
|
67 }
|
|
|
68
|
|
|
69 # Initialize class ...
|
|
|
70 sub _InitializeClass {
|
|
|
71 #Class name...
|
|
|
72 $ClassName = __PACKAGE__;
|
|
|
73
|
|
|
74 # Descriptor names...
|
|
|
75 @DescriptorNames = ('MolecularVolume');
|
|
|
76
|
|
|
77 # Initialize the data hash. It'll be loaded on demand later...
|
|
|
78 %VDWAtomRadiiAndVolumesDataMap = ();
|
|
|
79
|
|
|
80 }
|
|
|
81
|
|
|
82 # Get descriptor names as an array.
|
|
|
83 #
|
|
|
84 # This functionality can be either invoked as a class function or an
|
|
|
85 # object method.
|
|
|
86 #
|
|
|
87 sub GetDescriptorNames {
|
|
|
88 return @DescriptorNames;
|
|
|
89 }
|
|
|
90
|
|
|
91 # Initialize object data...
|
|
|
92 #
|
|
|
93 sub _InitializeMolecularVolumeDescriptors {
|
|
|
94 my($This) = @_;
|
|
|
95
|
|
|
96 # Type of MolecularDescriptor...
|
|
|
97 $This->{Type} = 'MolecularVolume';
|
|
|
98
|
|
|
99 # Intialize descriptor names and values...
|
|
|
100 $This->_InitializeDescriptorNamesAndValues(@DescriptorNames);
|
|
|
101
|
|
|
102 return $This;
|
|
|
103 }
|
|
|
104
|
|
|
105 # Initialize object properties...
|
|
|
106 #
|
|
|
107 sub _InitializeMolecularVolumeDescriptorsProperties {
|
|
|
108 my($This, %NamesAndValues) = @_;
|
|
|
109
|
|
|
110 my($Name, $Value, $MethodName);
|
|
|
111 while (($Name, $Value) = each %NamesAndValues) {
|
|
|
112 $MethodName = "Set${Name}";
|
|
|
113 $This->$MethodName($Value);
|
|
|
114 }
|
|
|
115
|
|
|
116 return $This;
|
|
|
117 }
|
|
|
118
|
|
|
119 # Get VDW atom data loaded from VDW atom radii and and volumes data file as
|
|
|
120 # a reference to hash with the following hash data format:
|
|
|
121 #
|
|
|
122 # @{$VDWAtomRadiiAndVolumesDataMap{AtomTypes}} - Array of all possible atom type symbols for all atoms
|
|
|
123 # @{$VDWAtomRadiiAndVolumesDataMap->{ColLabels}} - Array of column labels
|
|
|
124 # %{$VDWAtomRadiiAndVolumesDataMap->{DataCol<Num>}} - Hash keys pair: <DataCol<Num>, AtomType>
|
|
|
125 #
|
|
|
126 # This functionality can be either invoked as a class function or an
|
|
|
127 # object method.
|
|
|
128 #
|
|
|
129 sub GetVDWAtomRadiiAndVolumesData {
|
|
|
130
|
|
|
131 # Make sure data is loaded...
|
|
|
132 _CheckAndLoadVDWAtomRadiiAndVolumesData();
|
|
|
133
|
|
|
134 return \%VDWAtomRadiiAndVolumesDataMap;
|
|
|
135 }
|
|
|
136
|
|
|
137 # Calculate van der Waals molecular volume [ Ref 93 ] of a molecule using
|
|
|
138 # atomic and bonds contributions...
|
|
|
139 #
|
|
|
140 # van der Waals molecular volume (A**3/molecule) is defined as:
|
|
|
141 #
|
|
|
142 # vdwMolecularVolume = SumOfAtomicVDWVolumeContributions - 5.92 * NumOfBonds
|
|
|
143 # - 14.7 * NumOfAromaticRings - 3.8 * NumOfNonAromaticRings
|
|
|
144 #
|
|
|
145 # Methodology:
|
|
|
146 # . Add up van der Waals atom volumne of all atoms
|
|
|
147 # . Calculate molecular volume by correcting sum of atom volumes for num of
|
|
|
148 # bonds and rings
|
|
|
149 #
|
|
|
150 # Caveats:
|
|
|
151 # . All hydrogens must be added to molecule before calling GenerateDescriptors.
|
|
|
152 #
|
|
|
153 sub GenerateDescriptors {
|
|
|
154 my($This) = @_;
|
|
|
155
|
|
|
156 # Initialize descriptor values...
|
|
|
157 $This->_InitializeDescriptorValues();
|
|
|
158
|
|
|
159 # Check availability of molecule...
|
|
|
160 if (!$This->{Molecule}) {
|
|
|
161 carp "Warning: ${ClassName}->GenerateDescriptors: $This->{Type} molecular descriptors generation didn't succeed: Molecule data is not available: Molecule object hasn't been set...";
|
|
|
162 return undef;
|
|
|
163 }
|
|
|
164
|
|
|
165 # Calculate descriptor values...
|
|
|
166 if (!$This->_CalculateDescriptorValues()) {
|
|
|
167 carp "Warning: ${ClassName}->GenerateDescriptors: $This->{Type} molecular descriptors generation didn't succeed: Couldn't calculate MolecularVolume values: van der Waals atom volume data is not available for all atoms...";
|
|
|
168 return undef;
|
|
|
169 }
|
|
|
170
|
|
|
171 # Set final descriptor values...
|
|
|
172 $This->_SetFinalDescriptorValues();
|
|
|
173
|
|
|
174 return $This;
|
|
|
175 }
|
|
|
176
|
|
|
177 # Calculate MolecularVolume value...
|
|
|
178 #
|
|
|
179 sub _CalculateDescriptorValues {
|
|
|
180 my($This) = @_;
|
|
|
181 my($Atom, $AtomID, $AtomSymbol, $SumOfVDWAtomVolumes, $Molecule, $MolecularVolume, $NumOfBonds, $NumOfAromaticRings, $NumOfNonAromaticRings, $VDWAtomRadiiAndVolumesDataMapRef);
|
|
|
182
|
|
|
183 $MolecularVolume = 0;
|
|
|
184
|
|
|
185 $VDWAtomRadiiAndVolumesDataMapRef = $This->GetVDWAtomRadiiAndVolumesData();
|
|
|
186 $Molecule = $This->{Molecule};
|
|
|
187
|
|
|
188 # Calculate atom volumes contribution to molecular volume...
|
|
|
189 #
|
|
|
190 $SumOfVDWAtomVolumes = 0;
|
|
|
191
|
|
|
192 ATOM: for $Atom ($Molecule->GetAtoms()) {
|
|
|
193 $AtomID = $Atom->GetID();
|
|
|
194 $AtomSymbol = $Atom->GetAtomSymbol();
|
|
|
195
|
|
|
196 # Make sure van der Waals atom volume is available...
|
|
|
197 if (!exists $VDWAtomRadiiAndVolumesDataMap{DataCol3}{$AtomSymbol}) {
|
|
|
198 return undef;
|
|
|
199 }
|
|
|
200 $SumOfVDWAtomVolumes += $VDWAtomRadiiAndVolumesDataMapRef->{DataCol3}{$AtomSymbol};
|
|
|
201 }
|
|
|
202
|
|
|
203 $NumOfBonds = $Molecule->GetNumOfBonds();
|
|
|
204 $NumOfAromaticRings = $Molecule->GetNumOfAromaticRings();
|
|
|
205 $NumOfNonAromaticRings = $Molecule->GetNumOfRings() - $NumOfAromaticRings;
|
|
|
206
|
|
|
207 # Apply correction for bonds and rings...
|
|
|
208 $MolecularVolume = $SumOfVDWAtomVolumes - 5.92 * $NumOfBonds - 14.7 * $NumOfAromaticRings - 3.8 * $NumOfNonAromaticRings;
|
|
|
209
|
|
|
210 # Track the calculated values...
|
|
|
211 $This->{MolecularVolume} = MathUtil::round($MolecularVolume, 2);
|
|
|
212
|
|
|
213 return $This;
|
|
|
214 }
|
|
|
215
|
|
|
216 # Setup final descriptor values...
|
|
|
217 #
|
|
|
218 sub _SetFinalDescriptorValues {
|
|
|
219 my($This) = @_;
|
|
|
220
|
|
|
221 $This->{DescriptorsGenerated} = 1;
|
|
|
222
|
|
|
223 $This->SetDescriptorValues($This->{MolecularVolume});
|
|
|
224
|
|
|
225 return $This;
|
|
|
226 }
|
|
|
227
|
|
|
228 # Return a string containg data for MolecularVolumeDescriptors object...
|
|
|
229 #
|
|
|
230 sub StringifyMolecularVolumeDescriptors {
|
|
|
231 my($This) = @_;
|
|
|
232 my($MolecularVolumeDescriptorsString);
|
|
|
233
|
|
|
234 $MolecularVolumeDescriptorsString = "MolecularDescriptorType: $This->{Type}; " . $This->_StringifyDescriptorNamesAndValues();
|
|
|
235
|
|
|
236 return $MolecularVolumeDescriptorsString;
|
|
|
237 }
|
|
|
238
|
|
|
239 # Is it a MolecularVolumeDescriptors object?
|
|
|
240 sub _IsMolecularVolumeDescriptors {
|
|
|
241 my($Object) = @_;
|
|
|
242
|
|
|
243 return (Scalar::Util::blessed($Object) && $Object->isa($ClassName)) ? 1 : 0;
|
|
|
244 }
|
|
|
245
|
|
|
246 # Check and load van der Waals atom radii and volumes data...
|
|
|
247 #
|
|
|
248 sub _CheckAndLoadVDWAtomRadiiAndVolumesData {
|
|
|
249
|
|
|
250 # Is it already loaded?
|
|
|
251 if (exists $VDWAtomRadiiAndVolumesDataMap{AtomTypes}) {
|
|
|
252 return;
|
|
|
253 }
|
|
|
254
|
|
|
255 _LoadVDWAtomRadiiAndVolumesData();
|
|
|
256 }
|
|
|
257
|
|
|
258 # Initialize van der Waals atom radii and volumes data from the file...
|
|
|
259 #
|
|
|
260 # Format:
|
|
|
261 #
|
|
|
262 # "AtomTypeSymbol","VDWAtomRadius(A)","VDWAtomVolume(A**3)/molecule"
|
|
|
263 # "H","1.20","7.24"
|
|
|
264 # "He","1.40","11.49"
|
|
|
265 #
|
|
|
266 sub _LoadVDWAtomRadiiAndVolumesData {
|
|
|
267 my($VDWAtomDataFile, $MayaChemToolsLibDir);
|
|
|
268
|
|
|
269 $MayaChemToolsLibDir = FileUtil::GetMayaChemToolsLibDirName();
|
|
|
270
|
|
|
271 $VDWAtomDataFile = "$MayaChemToolsLibDir" . "/data/VDWAtomRadiiAndVolumes.csv";
|
|
|
272 if (! -e "$VDWAtomDataFile") {
|
|
|
273 croak "Error: MayaChemTools package file, $VDWAtomDataFile, is missing: Possible installation problems...";
|
|
|
274 }
|
|
|
275
|
|
|
276 %VDWAtomRadiiAndVolumesDataMap = ();
|
|
|
277 AtomTypes::AtomTypes::LoadAtomTypesData($VDWAtomDataFile, \%VDWAtomRadiiAndVolumesDataMap);
|
|
|
278 };
|
|
|
279
|
|
|
280 1;
|
|
|
281
|
|
|
282 __END__
|
|
|
283
|
|
|
284 =head1 NAME
|
|
|
285
|
|
|
286 MolecularVolumeDescriptors
|
|
|
287
|
|
|
288 =head1 SYNOPSIS
|
|
|
289
|
|
|
290 use MolecularDescriptors::MolecularVolumeDescriptors;
|
|
|
291
|
|
|
292 use MolecularDescriptors::MolecularVolumeDescriptors qw(:all);
|
|
|
293
|
|
|
294 =head1 DESCRIPTION
|
|
|
295
|
|
|
296 B<MolecularVolumeDescriptors> class provides the following methods:
|
|
|
297
|
|
|
298 new, GenerateDescriptors, GetDescriptorNames,
|
|
|
299 GetVDWAtomRadiiAndVolumesData, StringifyMolecularVolumeDescriptors
|
|
|
300
|
|
|
301 B<MolecularVolumeDescriptors> is derived from B<MolecularDescriptors> class which in turn
|
|
|
302 is derived from B<ObjectProperty> base class that provides methods not explicitly defined
|
|
|
303 in B<MolecularVolumeDescriptors>, B<MolecularDescriptors> or B<ObjectProperty> classes using Perl's
|
|
|
304 AUTOLOAD functionality. These methods are generated on-the-fly for a specified object property:
|
|
|
305
|
|
|
306 Set<PropertyName>(<PropertyValue>);
|
|
|
307 $PropertyValue = Get<PropertyName>();
|
|
|
308 Delete<PropertyName>();
|
|
|
309
|
|
|
310 van der Waals molecular volume [ Ref 93 ] (A**3/molecule) of a molecule is
|
|
|
311 calculated using atomic and bonds contributions along with adjustments for
|
|
|
312 aromatic and non-aromatic rings using the following equation:
|
|
|
313
|
|
|
314 vdwMolecularVolume = SumOfAtomicVDWVolumeContributions
|
|
|
315 - 5.92 * NumOfBonds
|
|
|
316 - 14.7 * NumOfAromaticRings
|
|
|
317 - 3.8 * NumOfNonAromaticRings
|
|
|
318
|
|
|
319 van der Waals atomic volume for atoms is taken from data file VDWAtomRadiiAndVolumes.csv
|
|
|
320 distributed with MayaChemTools. It contains van der Waals atom radii and atom and volumes
|
|
|
321 data for 38 elements; Table 2 [ Ref 93 ] contains data for only 15 elements. After converting
|
|
|
322 valid van der Waals atom radius data from pm (picometer) to A (Angstrom) available under column
|
|
|
323 name VanderWaalsRadius in PeriodicTableElementsData.csv data file, van der Waals atom volume
|
|
|
324 is calculated using: 4/3 * PI * (Radius ** 3). For elements specified in Table 2 [ Ref 93 ] -
|
|
|
325 H, B, C, N, O, F, Si, P, S, Cl, As, Se, Br, Te, I - the van der Waals atom radii and calculated
|
|
|
326 atom volumes match the values in the table.
|
|
|
327
|
|
|
328 =head2 METHODS
|
|
|
329
|
|
|
330 =over 4
|
|
|
331
|
|
|
332 =item B<new>
|
|
|
333
|
|
|
334 $NewMolecularVolumeDescriptors = new MolecularDescriptors::
|
|
|
335 MolecularVolumeDescriptors(
|
|
|
336 %NamesAndValues);
|
|
|
337
|
|
|
338 Using specified I<MolecularVolumeDescriptors> property names and values hash, B<new>
|
|
|
339 method creates a new object and returns a reference to newly created B<MolecularVolumeDescriptors>
|
|
|
340 object. By default, the following properties are initialized:
|
|
|
341
|
|
|
342 Molecule = ''
|
|
|
343 Type = 'MolecularVolume'
|
|
|
344 @DescriptorNames = ('MolecularVolume')
|
|
|
345 @DescriptorValues = ('None')
|
|
|
346
|
|
|
347 Examples:
|
|
|
348
|
|
|
349 $MolecularVolumeDescriptors = new MolecularDescriptors::
|
|
|
350 MolecularVolumeDescriptors();
|
|
|
351
|
|
|
352 $MolecularVolumeDescriptors->SetMolecule($Molecule);
|
|
|
353 $MolecularVolumeDescriptors->GenerateDescriptors();
|
|
|
354 print "MolecularVolumeDescriptors: $MolecularVolumeDescriptors\n";
|
|
|
355
|
|
|
356 =item B<GenerateDescriptors>
|
|
|
357
|
|
|
358 $MolecularVolumeDescriptors->GenerateDescriptors();
|
|
|
359
|
|
|
360 Calculate van der Waals molecular volume descriptor for a molecule and returns
|
|
|
361 I<MolecularVolumeDescriptors>.
|
|
|
362
|
|
|
363 =item B<GetDescriptorNames>
|
|
|
364
|
|
|
365 @DescriptorNames = $MolecularVolumeDescriptors->GetDescriptorNames();
|
|
|
366 @DescriptorNames = MolecularDescriptors::MolecularVolumeDescriptors::
|
|
|
367 GetDescriptorNames();
|
|
|
368
|
|
|
369 Returns all available descriptor names as an array.
|
|
|
370
|
|
|
371 =item B<GetVDWAtomRadiiAndVolumesData>
|
|
|
372
|
|
|
373 $VDWVolumeDataMapRef = $MolecularVolumeDescriptors->
|
|
|
374 GetVDWAtomRadiiAndVolumesData();
|
|
|
375 $VDWVolumeDataMapRef = MolecularDescriptors::MolecularVolumeDescriptors::
|
|
|
376 GetVDWAtomRadiiAndVolumesData();
|
|
|
377
|
|
|
378 Returns a hash reference to van der Waals atom symbols corresponding to atom types
|
|
|
379 and associated data loaded from VDWAtomRadiiAndVolumes.csv data file as a reference
|
|
|
380 to hash with the following hash data format:
|
|
|
381
|
|
|
382 @{$VDWVolumeDataMap{AtomTypes}} - Array of all possible atom
|
|
|
383 types for all atom symbols
|
|
|
384 @{$VDWVolumeDataMap->{ColLabels}} - Array of column labels
|
|
|
385 %{$VDWVolumeDataMap->{DataCol<Num>}} - Hash keys pair:
|
|
|
386 DataCol<Num>, AtomType
|
|
|
387
|
|
|
388 =item B<StringifyMolecularVolumeDescriptors>
|
|
|
389
|
|
|
390 $String = $MolecularVolumeDescriptors->
|
|
|
391 StringifyMolecularVolumeDescriptors();
|
|
|
392
|
|
|
393 Returns a string containing information about I<MolecularVolumeDescriptors> object.
|
|
|
394
|
|
|
395 =back
|
|
|
396
|
|
|
397 =head1 AUTHOR
|
|
|
398
|
|
|
399 Manish Sud <msud@san.rr.com>
|
|
|
400
|
|
|
401 =head1 SEE ALSO
|
|
|
402
|
|
|
403 MolecularDescriptors.pm, MolecularDescriptorsGenerator.pm
|
|
|
404
|
|
|
405 =head1 COPYRIGHT
|
|
|
406
|
|
|
407 Copyright (C) 2015 Manish Sud. All rights reserved.
|
|
|
408
|
|
|
409 This file is part of MayaChemTools.
|
|
|
410
|
|
|
411 MayaChemTools is free software; you can redistribute it and/or modify it under
|
|
|
412 the terms of the GNU Lesser General Public License as published by the Free
|
|
|
413 Software Foundation; either version 3 of the License, or (at your option)
|
|
|
414 any later version.
|
|
|
415
|
|
|
416 =cut
|
