|
0
|
1 <html>
|
|
|
2 <head>
|
|
|
3 <title>MayaChemTools:Code:Fingerprints::PathLengthFingerprints.pm</title>
|
|
|
4 <meta http-equiv="content-type" content="text/html;charset=utf-8">
|
|
|
5 <link rel="stylesheet" type="text/css" href="../../../css/MayaChemToolsCode.css">
|
|
|
6 </head>
|
|
|
7 <body leftmargin="20" rightmargin="20" topmargin="10" bottommargin="10">
|
|
|
8 <br/>
|
|
|
9 <center>
|
|
|
10 <a href="http://www.mayachemtools.org" title="MayaChemTools Home"><img src="../../../images/MayaChemToolsLogo.gif" border="0" alt="MayaChemTools"></a>
|
|
|
11 </center>
|
|
|
12 <br/>
|
|
|
13 <pre>
|
|
|
14 <a name="package-Fingerprints::PathLengthFingerprints-"></a> 1 <span class="k">package </span><span class="i">Fingerprints::PathLengthFingerprints</span><span class="sc">;</span>
|
|
|
15 2 <span class="c">#</span>
|
|
|
16 3 <span class="c"># $RCSfile: PathLengthFingerprints.pm,v $</span>
|
|
|
17 4 <span class="c"># $Date: 2015/02/28 20:48:54 $</span>
|
|
|
18 5 <span class="c"># $Revision: 1.39 $</span>
|
|
|
19 6 <span class="c">#</span>
|
|
|
20 7 <span class="c"># Author: Manish Sud <msud@san.rr.com></span>
|
|
|
21 8 <span class="c">#</span>
|
|
|
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span>
|
|
|
23 10 <span class="c">#</span>
|
|
|
24 11 <span class="c"># This file is part of MayaChemTools.</span>
|
|
|
25 12 <span class="c">#</span>
|
|
|
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span>
|
|
|
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span>
|
|
|
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span>
|
|
|
29 16 <span class="c"># later version.</span>
|
|
|
30 17 <span class="c">#</span>
|
|
|
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span>
|
|
|
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span>
|
|
|
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span>
|
|
|
34 21 <span class="c"># details.</span>
|
|
|
35 22 <span class="c">#</span>
|
|
|
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span>
|
|
|
37 24 <span class="c"># along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or</span>
|
|
|
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span>
|
|
|
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span>
|
|
|
40 27 <span class="c">#</span>
|
|
|
41 28
|
|
|
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span>
|
|
|
43 30 <span class="k">use</span> <span class="w">Carp</span><span class="sc">;</span>
|
|
|
44 31 <span class="k">use</span> <span class="w">Exporter</span><span class="sc">;</span>
|
|
|
45 32 <span class="k">use</span> <span class="w">TextUtil</span> <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
46 33 <span class="k">use</span> <span class="w">MathUtil</span> <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
47 34 <span class="k">use</span> <span class="w">Fingerprints::Fingerprints</span><span class="sc">;</span>
|
|
|
48 35 <span class="k">use</span> <span class="w">Molecule</span><span class="sc">;</span>
|
|
|
49 36 <span class="k">use</span> <span class="w">AtomTypes::AtomicInvariantsAtomTypes</span><span class="sc">;</span>
|
|
|
50 37 <span class="k">use</span> <span class="w">AtomTypes::DREIDINGAtomTypes</span><span class="sc">;</span>
|
|
|
51 38 <span class="k">use</span> <span class="w">AtomTypes::EStateAtomTypes</span><span class="sc">;</span>
|
|
|
52 39 <span class="k">use</span> <span class="w">AtomTypes::FunctionalClassAtomTypes</span><span class="sc">;</span>
|
|
|
53 40 <span class="k">use</span> <span class="w">AtomTypes::MMFF94AtomTypes</span><span class="sc">;</span>
|
|
|
54 41 <span class="k">use</span> <span class="w">AtomTypes::SLogPAtomTypes</span><span class="sc">;</span>
|
|
|
55 42 <span class="k">use</span> <span class="w">AtomTypes::SYBYLAtomTypes</span><span class="sc">;</span>
|
|
|
56 43 <span class="k">use</span> <span class="w">AtomTypes::TPSAAtomTypes</span><span class="sc">;</span>
|
|
|
57 44 <span class="k">use</span> <span class="w">AtomTypes::UFFAtomTypes</span><span class="sc">;</span>
|
|
|
58 45
|
|
|
59 46 <span class="k">use</span> <span class="w">vars</span> <span class="q">qw(@ISA @EXPORT @EXPORT_OK %EXPORT_TAGS)</span><span class="sc">;</span>
|
|
|
60 47
|
|
|
61 48 <span class="i">@ISA</span> = <span class="q">qw(Fingerprints::Fingerprints Exporter)</span><span class="sc">;</span>
|
|
|
62 49 <span class="i">@EXPORT</span> = <span class="q">qw()</span><span class="sc">;</span>
|
|
|
63 50 <span class="i">@EXPORT_OK</span> = <span class="q">qw()</span><span class="sc">;</span>
|
|
|
64 51
|
|
|
65 52 <span class="i">%EXPORT_TAGS</span> = <span class="s">(</span><span class="w">all</span> <span class="cm">=></span> <span class="s">[</span><span class="i">@EXPORT</span><span class="cm">,</span> <span class="i">@EXPORT_OK</span><span class="s">]</span><span class="s">)</span><span class="sc">;</span>
|
|
|
66 53
|
|
|
67 54 <span class="c"># Setup class variables...</span>
|
|
|
68 55 <span class="k">my</span><span class="s">(</span><span class="i">$ClassName</span><span class="s">)</span><span class="sc">;</span>
|
|
|
69 56 <span class="i">_InitializeClass</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
70 57
|
|
|
71 58 <span class="c"># Overload Perl functions...</span>
|
|
|
72 59 <span class="k">use</span> <span class="w">overload</span> <span class="q">'""'</span> <span class="cm">=></span> <span class="q">'StringifyPathLengthFingerprints'</span><span class="sc">;</span>
|
|
|
73 60
|
|
|
74 61 <span class="c"># Class constructor...</span>
|
|
|
75 <a name="new-"></a> 62 <span class="k">sub </span><span class="m">new</span> <span class="s">{</span>
|
|
|
76 63 <span class="k">my</span><span class="s">(</span><span class="i">$Class</span><span class="cm">,</span> <span class="i">%NamesAndValues</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
77 64
|
|
|
78 65 <span class="c"># Initialize object...</span>
|
|
|
79 66 <span class="k">my</span> <span class="i">$This</span> = <span class="i">$Class</span><span class="i">->SUPER::new</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
80 67 <span class="k">bless</span> <span class="i">$This</span><span class="cm">,</span> <span class="k">ref</span><span class="s">(</span><span class="i">$Class</span><span class="s">)</span> || <span class="i">$Class</span><span class="sc">;</span>
|
|
|
81 68 <span class="i">$This</span><span class="i">->_InitializePathLengthFingerprints</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
82 69
|
|
|
83 70 <span class="i">$This</span><span class="i">->_InitializePathLengthFingerprintsProperties</span><span class="s">(</span><span class="i">%NamesAndValues</span><span class="s">)</span><span class="sc">;</span>
|
|
|
84 71
|
|
|
85 72 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
86 73 <span class="s">}</span>
|
|
|
87 74
|
|
|
88 75 <span class="c"># Initialize object data...</span>
|
|
|
89 76 <span class="c">#</span>
|
|
|
90 <a name="_InitializePathLengthFingerprints-"></a> 77 <span class="k">sub </span><span class="m">_InitializePathLengthFingerprints</span> <span class="s">{</span>
|
|
|
91 78 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
92 79
|
|
|
93 80 <span class="c"># Type of fingerprint to generate...</span>
|
|
|
94 81 <span class="c">#</span>
|
|
|
95 82 <span class="c"># PathLengthBits - A bit vector indicating presence/absence of atom paths</span>
|
|
|
96 83 <span class="c"># PathLengthCount - A vector containing count of atom paths</span>
|
|
|
97 84 <span class="c">#</span>
|
|
|
98 85 <span class="i">$This</span>->{<span class="w">Type</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
|
99 86
|
|
|
100 87 <span class="c"># Type of vector: FingerprintsBitVector or FingerprintsVector</span>
|
|
|
101 88 <span class="i">$This</span>->{<span class="w">VectorType</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
|
102 89
|
|
|
103 90 <span class="c"># Set default mininum, maximum, and default size. Although any arbitrary size can</span>
|
|
|
104 91 <span class="c"># be specified, bit vector used to store bits work on a vector size which is</span>
|
|
|
105 92 <span class="c"># power of 2 and additonal bits are automatically added and cleared.</span>
|
|
|
106 93 <span class="c">#</span>
|
|
|
107 94 <span class="i">$This</span>->{<span class="w">Size</span>} = <span class="n">1024</span><span class="sc">;</span>
|
|
|
108 95
|
|
|
109 96 <span class="i">$This</span>->{<span class="w">MinSize</span>} = <span class="n">32</span><span class="sc">;</span>
|
|
|
110 97 <span class="i">$This</span>->{<span class="w">MaxSize</span>} = <span class="n">2</span>**<span class="n">32</span><span class="sc">;</span>
|
|
|
111 98
|
|
|
112 99 <span class="c"># Minimum and maximum path lengths to use for fingerprints generation...</span>
|
|
|
113 100 <span class="i">$This</span>->{<span class="w">MinLength</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
114 101 <span class="i">$This</span>->{<span class="w">MaxLength</span>} = <span class="n">8</span><span class="sc">;</span>
|
|
|
115 102
|
|
|
116 103 <span class="c"># Numner of bits to set for each atom path for FingerprintsBitVector...</span>
|
|
|
117 104 <span class="i">$This</span>->{<span class="w">NumOfBitsToSetPerPath</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
118 105
|
|
|
119 106 <span class="c"># Atom identifier type to use for path atoms during fingerprints generation...</span>
|
|
|
120 107 <span class="c">#</span>
|
|
|
121 108 <span class="c"># Currently supported values are: AtomicInvariantsAtomTypes, DREIDINGAtomTypes,</span>
|
|
|
122 109 <span class="c"># EStateAtomTypes, FunctionalClassAtomTypes, MMFF94AtomTypes, SLogPAtomTypes,</span>
|
|
|
123 110 <span class="c"># SYBYLAtomTypes, TPSAAtomTypes, UFFAtomTypes</span>
|
|
|
124 111 <span class="c">#</span>
|
|
|
125 112 <span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
|
126 113
|
|
|
127 114 <span class="c"># Atom types assigned to atoms...</span>
|
|
|
128 115 <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">AssignedAtomTypes</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
129 116
|
|
|
130 117 <span class="c"># For molecules containing rings, atom paths starting from each atom can be traversed in four</span>
|
|
|
131 118 <span class="c"># different ways:</span>
|
|
|
132 119 <span class="c">#</span>
|
|
|
133 120 <span class="c"># . Atom paths without any rings and sharing of bonds in traversed paths.</span>
|
|
|
134 121 <span class="c"># . Atom paths containing rings and without any sharing of bonds in traversed paths</span>
|
|
|
135 122 <span class="c"># . All possible atom paths without any rings and sharing of bonds in traversed paths</span>
|
|
|
136 123 <span class="c"># . All possible atom paths containing rings and with sharing of bonds in traversed paths.</span>
|
|
|
137 124 <span class="c">#</span>
|
|
|
138 125 <span class="c"># Atom path traversal is terminated at the last ring atom. For molecules containing no rings,</span>
|
|
|
139 126 <span class="c"># first two and last two types described above are equivalent.</span>
|
|
|
140 127 <span class="c">#</span>
|
|
|
141 128 <span class="c"># AllowSharedBonds and AllowRings variables allow generation of differen types of paths</span>
|
|
|
142 129 <span class="c"># to be used for fingerprints generation.</span>
|
|
|
143 130 <span class="c">#</span>
|
|
|
144 131 <span class="c"># In addition to atom symbols, bond symbols are also used to generate a string</span>
|
|
|
145 132 <span class="c"># for atom paths. These atom paths strings are hased to a 32 bit integer key which</span>
|
|
|
146 133 <span class="c"># in turn is used as a seed for a random number generation in range of 1 to fingerprint</span>
|
|
|
147 134 <span class="c"># size for setting corresponding bit in bit vector.</span>
|
|
|
148 135 <span class="c">#</span>
|
|
|
149 136 <span class="c"># UseBondSymbols variable allow generation of atom path strings and consequently fingerprints.</span>
|
|
|
150 137 <span class="c">#</span>
|
|
|
151 138 <span class="c"># Combination of AllowSharedBonds, AllowRings, and UseBondSymbols allow generation of</span>
|
|
|
152 139 <span class="c"># 8 different types of path length fingerprints:</span>
|
|
|
153 140 <span class="c">#</span>
|
|
|
154 141 <span class="c"># AllowSharedBonds AllowRings UseBondSymbols PathLengthFingerprintsType</span>
|
|
|
155 142 <span class="c">#</span>
|
|
|
156 143 <span class="c"># No No Yes AtomPathsNoCyclesWithBondSymbols</span>
|
|
|
157 144 <span class="c"># No Yes Yes AtomPathsWithCyclesWithBondSymbols</span>
|
|
|
158 145 <span class="c">#</span>
|
|
|
159 146 <span class="c"># Yes No Yes AllAtomPathsNoCyclesWithBondSymbols</span>
|
|
|
160 147 <span class="c"># Yes Yes Yes AllAtomPathsWithCyclesWithBondSymbols [ DEFAULT ]</span>
|
|
|
161 148 <span class="c">#</span>
|
|
|
162 149 <span class="c"># No No No AtomPathsNoCyclesNoBondSymbols</span>
|
|
|
163 150 <span class="c"># No Yes No AtomPathsWithCyclesNoBondSymbols</span>
|
|
|
164 151 <span class="c">#</span>
|
|
|
165 152 <span class="c"># Yes No No AllAtomPathsNoCyclesNoBondSymbols</span>
|
|
|
166 153 <span class="c"># Yes Yes No AllAtomPathsWithCyclesNoWithBondSymbols</span>
|
|
|
167 154 <span class="c">#</span>
|
|
|
168 155 <span class="c">#</span>
|
|
|
169 156
|
|
|
170 157 <span class="c"># By default, atom paths starting from atoms are allowed to share bonds already traversed...</span>
|
|
|
171 158 <span class="i">$This</span>->{<span class="w">AllowSharedBonds</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
172 159
|
|
|
173 160 <span class="c"># By default rings are included in paths...</span>
|
|
|
174 161 <span class="i">$This</span>->{<span class="w">AllowRings</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
175 162
|
|
|
176 163 <span class="c"># By default bond symbols are included in atom path strings...</span>
|
|
|
177 164 <span class="i">$This</span>->{<span class="w">UseBondSymbols</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
178 165
|
|
|
179 166 <span class="c"># By default only structurally unique atom paths are used for generation</span>
|
|
|
180 167 <span class="c"># atom path strings...</span>
|
|
|
181 168 <span class="i">$This</span>->{<span class="w">UseUniquePaths</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
182 169
|
|
|
183 170 <span class="c"># Random number generator to use during generation of fingerprints bit-vector</span>
|
|
|
184 171 <span class="c"># string: Perl CORE::rand or MayaChemTools MathUtil::random function.</span>
|
|
|
185 172 <span class="c">#</span>
|
|
|
186 173 <span class="c"># The random number generator implemented in MayaChemTools is a variant of</span>
|
|
|
187 174 <span class="c"># linear congruential generator (LCG) as described by Miller et al. [ Ref 120 ].</span>
|
|
|
188 175 <span class="c"># It is also referred to as Lehmer random number generator or Park-Miller</span>
|
|
|
189 176 <span class="c"># random number generator.</span>
|
|
|
190 177 <span class="c">#</span>
|
|
|
191 178 <span class="c"># Unlike Perl's core random number generator function rand, the random number</span>
|
|
|
192 179 <span class="c"># generator implemented in MayaChemTools, MathUtil::random, generates consistent</span>
|
|
|
193 180 <span class="c"># random values across different platformsfor a specific random seed and leads</span>
|
|
|
194 181 <span class="c"># to generation of portable fingerprints bit-vector strings.</span>
|
|
|
195 182 <span class="c">#</span>
|
|
|
196 183 <span class="i">$This</span>->{<span class="w">UsePerlCoreRandom</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
197 184
|
|
|
198 185 <span class="c"># Bond symbols to use during generation of atom path strings...</span>
|
|
|
199 186 <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">BondOrderToSymbol</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
200 187 <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">BondOrderToSymbol</span>}} = <span class="s">(</span><span class="q">'1'</span> <span class="cm">=></span> <span class="q">''</span><span class="cm">,</span> <span class="q">'1.5'</span> <span class="cm">=></span> <span class="q">':'</span><span class="cm">,</span> <span class="q">'2'</span> <span class="cm">=></span> <span class="q">'='</span><span class="cm">,</span> <span class="q">'3'</span> <span class="cm">=></span> <span class="q">'#'</span><span class="s">)</span><span class="sc">;</span>
|
|
|
201 188
|
|
|
202 189 <span class="c"># BondSymbols map to use for bonded atom IDs to use during atom path strings...</span>
|
|
|
203 190 <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">BondSymbols</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
204 191
|
|
|
205 192 <span class="c"># Path atom IDs to remove duplicate paths...</span>
|
|
|
206 193 <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">UniqueLinearAtomPathsIDs</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
207 194 <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">UniqueCyclicAtomPathsIDs</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
208 195
|
|
|
209 196 <span class="c"># Reference to all the atom paths upto specified path length...</span>
|
|
|
210 197 <span class="i">$This</span>->{<span class="w">AtomPathsRef</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
|
211 198
|
|
|
212 199 <span class="c"># Atom paths strings created using specified atom types and bond symbols...</span>
|
|
|
213 200 <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
214 201 <span class="s">}</span>
|
|
|
215 202
|
|
|
216 203 <span class="c"># Initialize class ...</span>
|
|
|
217 <a name="_InitializeClass-"></a> 204 <span class="k">sub </span><span class="m">_InitializeClass</span> <span class="s">{</span>
|
|
|
218 205 <span class="c">#Class name...</span>
|
|
|
219 206 <span class="i">$ClassName</span> = <span class="w">__PACKAGE__</span><span class="sc">;</span>
|
|
|
220 207 <span class="s">}</span>
|
|
|
221 208
|
|
|
222 209 <span class="c"># Initialize object properties....</span>
|
|
|
223 <a name="_InitializePathLengthFingerprintsProperties-"></a> 210 <span class="k">sub </span><span class="m">_InitializePathLengthFingerprintsProperties</span> <span class="s">{</span>
|
|
|
224 211 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">%NamesAndValues</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
225 212
|
|
|
226 213 <span class="k">my</span><span class="s">(</span><span class="i">$Name</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">$MethodName</span><span class="s">)</span><span class="sc">;</span>
|
|
|
227 214 <span class="k">while</span> <span class="s">(</span><span class="s">(</span><span class="i">$Name</span><span class="cm">,</span> <span class="i">$Value</span><span class="s">)</span> = <span class="k">each</span> <span class="i">%NamesAndValues</span><span class="s">)</span> <span class="s">{</span>
|
|
|
228 215 <span class="i">$MethodName</span> = <span class="q">"Set${Name}"</span><span class="sc">;</span>
|
|
|
229 216 <span class="i">$This</span><span class="i">->$MethodName</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span><span class="sc">;</span>
|
|
|
230 217 <span class="s">}</span>
|
|
|
231 218
|
|
|
232 219 <span class="c"># Make sure molecule object was specified...</span>
|
|
|
233 220 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$NamesAndValues</span>{<span class="w">Molecule</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
234 221 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->New: Object can't be instantiated without specifying molecule..."</span><span class="sc">;</span>
|
|
|
235 222 <span class="s">}</span>
|
|
|
236 223
|
|
|
237 224 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$NamesAndValues</span>{<span class="w">Type</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
238 225 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->New: Object can't be instantiated without specifying Type..."</span><span class="sc">;</span>
|
|
|
239 226 <span class="s">}</span>
|
|
|
240 227
|
|
|
241 228 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$NamesAndValues</span>{<span class="w">AtomIdentifierType</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
242 229 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->New: Object can't be instantiated without specifying AtomIdentifierType..."</span><span class="sc">;</span>
|
|
|
243 230 <span class="s">}</span>
|
|
|
244 231
|
|
|
245 232 <span class="c"># Make sure it's power of 2...</span>
|
|
|
246 233 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$NamesAndValues</span>{<span class="w">Size</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
247 234 <span class="k">if</span> <span class="s">(</span>!<span class="i">TextUtil::IsNumberPowerOfNumber</span><span class="s">(</span><span class="i">$NamesAndValues</span>{<span class="w">Size</span>}<span class="cm">,</span> <span class="n">2</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
248 235 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->New: Specified size value, $NamesAndValues{Size}, must be power of 2..."</span><span class="sc">;</span>
|
|
|
249 236 <span class="s">}</span>
|
|
|
250 237 <span class="s">}</span>
|
|
|
251 238
|
|
|
252 239 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">Type</span>} =~ <span class="q">/^PathLengthBits$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
253 240 <span class="i">$This</span><span class="i">->_InitializePathLengthBits</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
254 241 <span class="s">}</span>
|
|
|
255 242 <span class="k">elsif</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">Type</span>} =~ <span class="q">/^PathLengthCount$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
256 243 <span class="i">$This</span><span class="i">->_InitializePathLengthCount</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
257 244 <span class="s">}</span>
|
|
|
258 245 <span class="k">else</span> <span class="s">{</span>
|
|
|
259 246 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->_InitializePathLengthFingerprintsProperties: Unknown PathLength type: $This->{Type}; Supported PathLength type : PathLengthBits or PathLengthCount......"</span><span class="sc">;</span>
|
|
|
260 247 <span class="s">}</span>
|
|
|
261 248
|
|
|
262 249 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
263 250 <span class="s">}</span>
|
|
|
264 251
|
|
|
265 252 <span class="c"># Initialize PathLength bits...</span>
|
|
|
266 253 <span class="c">#</span>
|
|
|
267 <a name="_InitializePathLengthBits-"></a> 254 <span class="k">sub </span><span class="m">_InitializePathLengthBits</span> <span class="s">{</span>
|
|
|
268 255 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
269 256
|
|
|
270 257 <span class="c"># Vector type...</span>
|
|
|
271 258 <span class="i">$This</span>->{<span class="w">VectorType</span>} = <span class="q">'FingerprintsBitVector'</span><span class="sc">;</span>
|
|
|
272 259
|
|
|
273 260 <span class="i">$This</span><span class="i">->_InitializeFingerprintsBitVector</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
274 261
|
|
|
275 262 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
276 263 <span class="s">}</span>
|
|
|
277 264
|
|
|
278 265 <span class="c"># Initialize PathLength key count...</span>
|
|
|
279 266 <span class="c">#</span>
|
|
|
280 <a name="_InitializePathLengthCount-"></a> 267 <span class="k">sub </span><span class="m">_InitializePathLengthCount</span> <span class="s">{</span>
|
|
|
281 268 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
282 269
|
|
|
283 270 <span class="c"># Vector type and type of values...</span>
|
|
|
284 271 <span class="i">$This</span>->{<span class="w">VectorType</span>} = <span class="q">'FingerprintsVector'</span><span class="sc">;</span>
|
|
|
285 272 <span class="i">$This</span>->{<span class="w">FingerprintsVectorType</span>} = <span class="q">'NumericalValues'</span><span class="sc">;</span>
|
|
|
286 273
|
|
|
287 274 <span class="i">$This</span><span class="i">->_InitializeFingerprintsVector</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
288 275
|
|
|
289 276 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
290 277 <span class="s">}</span>
|
|
|
291 278
|
|
|
292 279 <span class="c"># Set type...</span>
|
|
|
293 280 <span class="c">#</span>
|
|
|
294 <a name="SetType-"></a> 281 <span class="k">sub </span><span class="m">SetType</span> <span class="s">{</span>
|
|
|
295 282 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$Type</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
296 283
|
|
|
297 284 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">Type</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
298 285 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->SetType: Can't change type: It's already set..."</span><span class="sc">;</span>
|
|
|
299 286 <span class="s">}</span>
|
|
|
300 287
|
|
|
301 288 <span class="k">if</span> <span class="s">(</span><span class="i">$Type</span> =~ <span class="q">/^PathLengthBits$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
302 289 <span class="i">$This</span>->{<span class="w">Type</span>} = <span class="q">'PathLengthBits'</span><span class="sc">;</span><span class="sc">;</span>
|
|
|
303 290 <span class="s">}</span>
|
|
|
304 291 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Type</span> =~ <span class="q">/^PathLengthCount$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
305 292 <span class="i">$This</span>->{<span class="w">Type</span>} = <span class="q">'PathLengthCount'</span><span class="sc">;</span><span class="sc">;</span>
|
|
|
306 293 <span class="s">}</span>
|
|
|
307 294 <span class="k">else</span> <span class="s">{</span>
|
|
|
308 295 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->SetType: Unknown PathLength keys: $Type; Supported PathLength types: PathLengthBits or PathLengthCount..."</span><span class="sc">;</span>
|
|
|
309 296 <span class="s">}</span>
|
|
|
310 297 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
311 298 <span class="s">}</span>
|
|
|
312 299
|
|
|
313 300 <span class="c"># Disable vector type change...</span>
|
|
|
314 301 <span class="c">#</span>
|
|
|
315 <a name="SetVectorType-"></a> 302 <span class="k">sub </span><span class="m">SetVectorType</span> <span class="s">{</span>
|
|
|
316 303 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$Type</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
317 304
|
|
|
318 305 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->SetVectorType: Can't change vector type..."</span><span class="sc">;</span>
|
|
|
319 306
|
|
|
320 307 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
321 308 <span class="s">}</span>
|
|
|
322 309
|
|
|
323 310 <span class="c"># Disable vector type change...</span>
|
|
|
324 311 <span class="c">#</span>
|
|
|
325 <a name="SetFingerprintsVectorType-"></a> 312 <span class="k">sub </span><span class="m">SetFingerprintsVectorType</span> <span class="s">{</span>
|
|
|
326 313 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$Type</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
327 314
|
|
|
328 315 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->SetFingerprintsVectorType: Can't change fingerprints vector type..."</span><span class="sc">;</span>
|
|
|
329 316
|
|
|
330 317 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
331 318 <span class="s">}</span>
|
|
|
332 319
|
|
|
333 320 <span class="c"># Set atom identifier type to use for path length atom identifiers...</span>
|
|
|
334 321 <span class="c">#</span>
|
|
|
335 <a name="SetAtomIdentifierType-"></a> 322 <span class="k">sub </span><span class="m">SetAtomIdentifierType</span> <span class="s">{</span>
|
|
|
336 323 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$IdentifierType</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
337 324
|
|
|
338 325 <span class="k">if</span> <span class="s">(</span><span class="i">$IdentifierType</span> !~ <span class="q">/^(AtomicInvariantsAtomTypes|DREIDINGAtomTypes|EStateAtomTypes|FunctionalClassAtomTypes|MMFF94AtomTypes|SLogPAtomTypes|SYBYLAtomTypes|TPSAAtomTypes|UFFAtomTypes)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
339 326 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->SetAtomIdentifierType: Specified value, $IdentifierType, for AtomIdentifierType is not vaild. Supported types in current release of MayaChemTools: AtomicInvariantsAtomTypes, DREIDINGAtomTypes, EStateAtomTypes, FunctionalClassAtomTypes, MMFF94AtomTypes, SLogPAtomTypes, SYBYLAtomTypes, TPSAAtomTypes, and UFFAtomTypes."</span><span class="sc">;</span>
|
|
|
340 327 <span class="s">}</span>
|
|
|
341 328
|
|
|
342 329 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
343 330 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->SetAtomIdentifierType: Can't change atom identifier type: It's already set..."</span><span class="sc">;</span>
|
|
|
344 331 <span class="s">}</span>
|
|
|
345 332
|
|
|
346 333 <span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} = <span class="i">$IdentifierType</span><span class="sc">;</span>
|
|
|
347 334
|
|
|
348 335 <span class="c"># Initialize atom identifier type information...</span>
|
|
|
349 336 <span class="i">$This</span><span class="i">->_InitializeAtomIdentifierTypeInformation</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
350 337
|
|
|
351 338 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
352 339 <span class="s">}</span>
|
|
|
353 340
|
|
|
354 341 <span class="c"># Set minimum path length...</span>
|
|
|
355 342 <span class="c">#</span>
|
|
|
356 <a name="SetMinLength-"></a> 343 <span class="k">sub </span><span class="m">SetMinLength</span> <span class="s">{</span>
|
|
|
357 344 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$Value</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
358 345
|
|
|
359 346 <span class="k">if</span> <span class="s">(</span>!<span class="i">TextUtil::IsPositiveInteger</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
360 347 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->SetMinLength: MinLength value, $Value, is not valid: It must be a positive integer..."</span><span class="sc">;</span>
|
|
|
361 348 <span class="s">}</span>
|
|
|
362 349 <span class="i">$This</span>->{<span class="w">MinLength</span>} = <span class="i">$Value</span><span class="sc">;</span>
|
|
|
363 350
|
|
|
364 351 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
365 352 <span class="s">}</span>
|
|
|
366 353
|
|
|
367 354 <span class="c"># Set maximum path length...</span>
|
|
|
368 355 <span class="c">#</span>
|
|
|
369 <a name="SetMaxLength-"></a> 356 <span class="k">sub </span><span class="m">SetMaxLength</span> <span class="s">{</span>
|
|
|
370 357 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$Value</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
371 358
|
|
|
372 359 <span class="k">if</span> <span class="s">(</span>!<span class="i">TextUtil::IsPositiveInteger</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
373 360 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->SetMaxLength: MaxLength value, $Value, is not valid: It must be a positive integer..."</span><span class="sc">;</span>
|
|
|
374 361 <span class="s">}</span>
|
|
|
375 362 <span class="i">$This</span>->{<span class="w">MaxLength</span>} = <span class="i">$Value</span><span class="sc">;</span>
|
|
|
376 363
|
|
|
377 364 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
378 365 <span class="s">}</span>
|
|
|
379 366
|
|
|
380 367 <span class="c"># Set number of bits to set for each path...</span>
|
|
|
381 368 <span class="c">#</span>
|
|
|
382 <a name="SetNumOfBitsToSetPerPath-"></a> 369 <span class="k">sub </span><span class="m">SetNumOfBitsToSetPerPath</span> <span class="s">{</span>
|
|
|
383 370 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$Value</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
384 371
|
|
|
385 372 <span class="k">if</span> <span class="s">(</span>!<span class="i">TextUtil::IsPositiveInteger</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
386 373 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->SetNumOfBitsToSetPerPath: NumOfBitsToSetPerPath value, $Value, is not valid: It must be a positive integer..."</span><span class="sc">;</span>
|
|
|
387 374 <span class="s">}</span>
|
|
|
388 375 <span class="i">$This</span>->{<span class="w">NumOfBitsToSetPerPath</span>} = <span class="i">$Value</span><span class="sc">;</span>
|
|
|
389 376
|
|
|
390 377 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
391 378 <span class="s">}</span>
|
|
|
392 379
|
|
|
393 380 <span class="c"># Generate fingerprints description...</span>
|
|
|
394 381 <span class="c">#</span>
|
|
|
395 <a name="GetDescription-"></a> 382 <span class="k">sub </span><span class="m">GetDescription</span> <span class="s">{</span>
|
|
|
396 383 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
397 384
|
|
|
398 385 <span class="c"># Is description explicity set?</span>
|
|
|
399 386 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$This</span>->{<span class="w">Description</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
400 387 <span class="k">return</span> <span class="i">$This</span>->{<span class="w">Description</span>}<span class="sc">;</span>
|
|
|
401 388 <span class="s">}</span>
|
|
|
402 389
|
|
|
403 390 <span class="c"># Generate fingerprints description...</span>
|
|
|
404 391
|
|
|
405 392 <span class="k">return</span> <span class="q">"$This->{Type}:$This->{AtomIdentifierType}:MinLength$This->{MinLength}:MaxLength$This->{MaxLength}"</span><span class="sc">;</span>
|
|
|
406 393 <span class="s">}</span>
|
|
|
407 394
|
|
|
408 395 <span class="c"># Generate path length fingerprints...</span>
|
|
|
409 396 <span class="c">#</span>
|
|
|
410 <a name="GenerateFingerprints-"></a> 397 <span class="k">sub </span><span class="m">GenerateFingerprints</span> <span class="s">{</span>
|
|
|
411 398 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
412 399
|
|
|
413 400 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">MinLength</span>} > <span class="i">$This</span>->{<span class="w">MaxLength</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
414 401 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->GenerateFingerprints: No fingerpritns generated: MinLength, $This->{MinLength}, must be <= MaxLength, $This->{MaxLength}..."</span><span class="sc">;</span>
|
|
|
415 402 <span class="s">}</span>
|
|
|
416 403
|
|
|
417 404 <span class="c"># Cache appropriate molecule data...</span>
|
|
|
418 405 <span class="i">$This</span><span class="i">->_SetupMoleculeDataCache</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
419 406
|
|
|
420 407 <span class="c"># Assign atom types to all atoms...</span>
|
|
|
421 408 <span class="k">if</span> <span class="s">(</span>!<span class="i">$This</span><span class="i">->_AssignAtomTypes</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
422 409 <span class="w">carp</span> <span class="q">"Warning: ${ClassName}->GenerateFingerprints: $This->{AtomIdentifierType} fingerprints generation didn't succeed: Couldn't assign valid $This->{AtomIdentifierType} to all atoms..."</span><span class="sc">;</span>
|
|
|
423 410 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
424 411 <span class="s">}</span>
|
|
|
425 412
|
|
|
426 413 <span class="c"># Setup bond symbol map...</span>
|
|
|
427 414 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">UseBondSymbols</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
428 415 <span class="i">$This</span><span class="i">->_InitializeBondSymbols</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
429 416 <span class="s">}</span>
|
|
|
430 417
|
|
|
431 418 <span class="c"># Generate appropriate atom paths...</span>
|
|
|
432 419 <span class="i">$This</span><span class="i">->_GenerateAtomPathsUpToMaxLength</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
433 420
|
|
|
434 421 <span class="c"># Initialize atom path strings...</span>
|
|
|
435 422 <span class="i">$This</span><span class="i">->_InitializeAtomPathsStrings</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
436 423
|
|
|
437 424 <span class="c"># Generate appropriate atom path strings for unique atom paths...</span>
|
|
|
438 425 <span class="i">$This</span><span class="i">->_GenerateAtomPathsStrings</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
439 426
|
|
|
440 427 <span class="c"># Set final fingerprints...</span>
|
|
|
441 428 <span class="i">$This</span><span class="i">->_SetFinalFingerprints</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
442 429
|
|
|
443 430 <span class="c"># Clear cached molecule data...</span>
|
|
|
444 431 <span class="i">$This</span><span class="i">->_ClearMoleculeDataCache</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
445 432
|
|
|
446 433 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
447 434 <span class="s">}</span>
|
|
|
448 435
|
|
|
449 436 <span class="c"># Assign appropriate atom types to all atoms...</span>
|
|
|
450 437 <span class="c">#</span>
|
|
|
451 <a name="_AssignAtomTypes-"></a> 438 <span class="k">sub </span><span class="m">_AssignAtomTypes</span> <span class="s">{</span>
|
|
|
452 439 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
453 440 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedAtomTypes</span><span class="cm">,</span> <span class="i">$Atom</span><span class="cm">,</span> <span class="i">$AtomID</span><span class="cm">,</span> <span class="i">$IgnoreHydrogens</span><span class="s">)</span><span class="sc">;</span>
|
|
|
454 441
|
|
|
455 442 <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">AssignedAtomTypes</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
456 443 <span class="i">$IgnoreHydrogens</span> = <span class="n">0</span><span class="sc">;</span>
|
|
|
457 444
|
|
|
458 445 <span class="i">$SpecifiedAtomTypes</span> = <span class="k">undef</span><span class="sc">;</span>
|
|
|
459 446
|
|
|
460 447 <span class="j">IDENTIFIERTYPE:</span> <span class="s">{</span>
|
|
|
461 448 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^AtomicInvariantsAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
462 449 <span class="i">$SpecifiedAtomTypes</span> = <span class="i">new</span> <span class="i">AtomTypes::AtomicInvariantsAtomTypes</span><span class="s">(</span><span class="q">'Molecule'</span> <span class="cm">=></span> <span class="i">$This</span>->{<span class="w">Molecule</span>}<span class="cm">,</span> <span class="q">'IgnoreHydrogens'</span> <span class="cm">=></span> <span class="i">$IgnoreHydrogens</span><span class="cm">,</span> <span class="q">'AtomicInvariantsToUse'</span> <span class="cm">=></span> <span class="i">$This</span>->{<span class="w">AtomicInvariantsToUse</span>}<span class="s">)</span><span class="sc">;</span>
|
|
|
463 450 <span class="k">last</span> <span class="j">IDENTIFIERTYPE</span><span class="sc">;</span>
|
|
|
464 451 <span class="s">}</span>
|
|
|
465 452
|
|
|
466 453 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^DREIDINGAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
467 454 <span class="i">$SpecifiedAtomTypes</span> = <span class="i">new</span> <span class="i">AtomTypes::DREIDINGAtomTypes</span><span class="s">(</span><span class="q">'Molecule'</span> <span class="cm">=></span> <span class="i">$This</span>->{<span class="w">Molecule</span>}<span class="cm">,</span> <span class="q">'IgnoreHydrogens'</span> <span class="cm">=></span> <span class="i">$IgnoreHydrogens</span><span class="s">)</span><span class="sc">;</span>
|
|
|
468 455 <span class="k">last</span> <span class="j">IDENTIFIERTYPE</span><span class="sc">;</span>
|
|
|
469 456 <span class="s">}</span>
|
|
|
470 457
|
|
|
471 458 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^EStateAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
472 459 <span class="i">$SpecifiedAtomTypes</span> = <span class="i">new</span> <span class="i">AtomTypes::EStateAtomTypes</span><span class="s">(</span><span class="q">'Molecule'</span> <span class="cm">=></span> <span class="i">$This</span>->{<span class="w">Molecule</span>}<span class="cm">,</span> <span class="q">'IgnoreHydrogens'</span> <span class="cm">=></span> <span class="i">$IgnoreHydrogens</span><span class="s">)</span><span class="sc">;</span>
|
|
|
473 460 <span class="k">last</span> <span class="j">IDENTIFIERTYPE</span><span class="sc">;</span>
|
|
|
474 461 <span class="s">}</span>
|
|
|
475 462
|
|
|
476 463 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^FunctionalClassAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
477 464 <span class="i">$SpecifiedAtomTypes</span> = <span class="i">new</span> <span class="i">AtomTypes::FunctionalClassAtomTypes</span><span class="s">(</span><span class="q">'Molecule'</span> <span class="cm">=></span> <span class="i">$This</span>->{<span class="w">Molecule</span>}<span class="cm">,</span> <span class="q">'IgnoreHydrogens'</span> <span class="cm">=></span> <span class="i">$IgnoreHydrogens</span><span class="cm">,</span> <span class="q">'FunctionalClassesToUse'</span> <span class="cm">=></span> <span class="i">$This</span>->{<span class="w">FunctionalClassesToUse</span>}<span class="s">)</span><span class="sc">;</span>
|
|
|
478 465 <span class="k">last</span> <span class="j">IDENTIFIERTYPE</span><span class="sc">;</span>
|
|
|
479 466 <span class="s">}</span>
|
|
|
480 467
|
|
|
481 468 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^MMFF94AtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
482 469 <span class="i">$SpecifiedAtomTypes</span> = <span class="i">new</span> <span class="i">AtomTypes::MMFF94AtomTypes</span><span class="s">(</span><span class="q">'Molecule'</span> <span class="cm">=></span> <span class="i">$This</span>->{<span class="w">Molecule</span>}<span class="cm">,</span> <span class="q">'IgnoreHydrogens'</span> <span class="cm">=></span> <span class="i">$IgnoreHydrogens</span><span class="s">)</span><span class="sc">;</span>
|
|
|
483 470 <span class="k">last</span> <span class="j">IDENTIFIERTYPE</span><span class="sc">;</span>
|
|
|
484 471 <span class="s">}</span>
|
|
|
485 472
|
|
|
486 473 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^SLogPAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
487 474 <span class="i">$SpecifiedAtomTypes</span> = <span class="i">new</span> <span class="i">AtomTypes::SLogPAtomTypes</span><span class="s">(</span><span class="q">'Molecule'</span> <span class="cm">=></span> <span class="i">$This</span>->{<span class="w">Molecule</span>}<span class="cm">,</span> <span class="q">'IgnoreHydrogens'</span> <span class="cm">=></span> <span class="i">$IgnoreHydrogens</span><span class="s">)</span><span class="sc">;</span>
|
|
|
488 475 <span class="k">last</span> <span class="j">IDENTIFIERTYPE</span><span class="sc">;</span>
|
|
|
489 476 <span class="s">}</span>
|
|
|
490 477 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^SYBYLAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
491 478 <span class="i">$SpecifiedAtomTypes</span> = <span class="i">new</span> <span class="i">AtomTypes::SYBYLAtomTypes</span><span class="s">(</span><span class="q">'Molecule'</span> <span class="cm">=></span> <span class="i">$This</span>->{<span class="w">Molecule</span>}<span class="cm">,</span> <span class="q">'IgnoreHydrogens'</span> <span class="cm">=></span> <span class="i">$IgnoreHydrogens</span><span class="s">)</span><span class="sc">;</span>
|
|
|
492 479 <span class="k">last</span> <span class="j">IDENTIFIERTYPE</span><span class="sc">;</span>
|
|
|
493 480 <span class="s">}</span>
|
|
|
494 481
|
|
|
495 482 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^TPSAAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
496 483 <span class="i">$SpecifiedAtomTypes</span> = <span class="i">new</span> <span class="i">AtomTypes::TPSAAtomTypes</span><span class="s">(</span><span class="q">'Molecule'</span> <span class="cm">=></span> <span class="i">$This</span>->{<span class="w">Molecule</span>}<span class="cm">,</span> <span class="q">'IgnorePhosphorus'</span> <span class="cm">=></span> <span class="n">0</span><span class="cm">,</span> <span class="q">'IgnoreSulfur'</span> <span class="cm">=></span> <span class="n">0</span><span class="s">)</span><span class="sc">;</span>
|
|
|
497 484 <span class="k">last</span> <span class="j">IDENTIFIERTYPE</span><span class="sc">;</span>
|
|
|
498 485 <span class="s">}</span>
|
|
|
499 486
|
|
|
500 487 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^UFFAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
501 488 <span class="i">$SpecifiedAtomTypes</span> = <span class="i">new</span> <span class="i">AtomTypes::UFFAtomTypes</span><span class="s">(</span><span class="q">'Molecule'</span> <span class="cm">=></span> <span class="i">$This</span>->{<span class="w">Molecule</span>}<span class="cm">,</span> <span class="q">'IgnoreHydrogens'</span> <span class="cm">=></span> <span class="i">$IgnoreHydrogens</span><span class="s">)</span><span class="sc">;</span>
|
|
|
502 489 <span class="k">last</span> <span class="j">IDENTIFIERTYPE</span><span class="sc">;</span>
|
|
|
503 490 <span class="s">}</span>
|
|
|
504 491
|
|
|
505 492 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->_AssignAtomTypes: Unknown atom indentifier type $This->{AtomIdentifierType}..."</span><span class="sc">;</span>
|
|
|
506 493 <span class="s">}</span>
|
|
|
507 494
|
|
|
508 495 <span class="c"># Assign atom types...</span>
|
|
|
509 496 <span class="i">$SpecifiedAtomTypes</span><span class="i">->AssignAtomTypes</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
510 497
|
|
|
511 498 <span class="c"># Make sure atom types assignment is successful...</span>
|
|
|
512 499 <span class="k">if</span> <span class="s">(</span>!<span class="i">$SpecifiedAtomTypes</span><span class="i">->IsAtomTypesAssignmentSuccessful</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
513 500 <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span>
|
|
|
514 501 <span class="s">}</span>
|
|
|
515 502
|
|
|
516 503 <span class="c"># Collect assigned atom types...</span>
|
|
|
517 504 <span class="j">ATOM:</span> <span class="k">for</span> <span class="i">$Atom</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$This</span>->{<span class="w">Atoms</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
518 505 <span class="i">$AtomID</span> = <span class="i">$Atom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
519 506 <span class="i">$This</span>->{<span class="w">AssignedAtomTypes</span>}{<span class="i">$AtomID</span>} = <span class="i">$SpecifiedAtomTypes</span><span class="i">->GetAtomType</span><span class="s">(</span><span class="i">$Atom</span><span class="s">)</span><span class="sc">;</span>
|
|
|
520 507 <span class="s">}</span>
|
|
|
521 508
|
|
|
522 509 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
523 510 <span class="s">}</span>
|
|
|
524 511
|
|
|
525 512 <span class="c"># Setup bond symbol map for atoms to speed up generation of path length identifiers</span>
|
|
|
526 513 <span class="c"># during fingerprints generation...</span>
|
|
|
527 514 <span class="c">#</span>
|
|
|
528 <a name="_InitializeBondSymbols-"></a> 515 <span class="k">sub </span><span class="m">_InitializeBondSymbols</span> <span class="s">{</span>
|
|
|
529 516 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
530 517 <span class="k">my</span><span class="s">(</span><span class="i">$Atom1</span><span class="cm">,</span> <span class="i">$Atom2</span><span class="cm">,</span> <span class="i">$AtomID1</span><span class="cm">,</span> <span class="i">$AtomID2</span><span class="cm">,</span> <span class="i">$Bond</span><span class="cm">,</span> <span class="i">$BondSymbol</span><span class="cm">,</span> <span class="i">$BondOrder</span><span class="s">)</span><span class="sc">;</span>
|
|
|
531 518
|
|
|
532 519 <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">BondSymbols</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
533 520
|
|
|
534 521 <span class="k">if</span> <span class="s">(</span>!<span class="i">$This</span>->{<span class="w">UseBondSymbols</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
535 522 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
536 523 <span class="s">}</span>
|
|
|
537 524
|
|
|
538 525 <span class="k">for</span> <span class="i">$Bond</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">Molecule</span>}<span class="i">->GetBonds</span><span class="s">(</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
539 526 <span class="i">$BondOrder</span> = <span class="i">$Bond</span><span class="i">->GetBondOrder</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
540 527 <span class="i">$BondSymbol</span> = <span class="i">$Bond</span><span class="i">->IsAromatic</span><span class="s">(</span><span class="s">)</span> ? <span class="q">':'</span> <span class="co">:</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$This</span>->{<span class="w">BondOrderToSymbol</span>}{<span class="i">$BondOrder</span>}<span class="s">)</span> ? <span class="i">$This</span>->{<span class="w">BondOrderToSymbol</span>}{<span class="i">$BondOrder</span>} <span class="co">:</span> <span class="i">$BondOrder</span><span class="s">)</span><span class="sc">;</span>
|
|
|
541 528 <span class="s">(</span><span class="i">$Atom1</span><span class="cm">,</span> <span class="i">$Atom2</span><span class="s">)</span> = <span class="i">$Bond</span><span class="i">->GetAtoms</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
542 529 <span class="i">$AtomID1</span> = <span class="i">$Atom1</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> <span class="i">$AtomID2</span> = <span class="i">$Atom2</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
543 530 <span class="k">if</span> <span class="s">(</span><span class="i">$AtomID1</span> > <span class="i">$AtomID2</span><span class="s">)</span> <span class="s">{</span>
|
|
|
544 531 <span class="s">(</span><span class="i">$AtomID1</span><span class="cm">,</span> <span class="i">$AtomID2</span><span class="s">)</span> = <span class="s">(</span><span class="i">$AtomID2</span><span class="cm">,</span> <span class="i">$AtomID1</span><span class="s">)</span><span class="sc">;</span>
|
|
|
545 532 <span class="s">}</span>
|
|
|
546 533
|
|
|
547 534 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$This</span>->{<span class="w">BondSymbols</span>}{<span class="i">$AtomID1</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
548 535 <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">BondSymbols</span>}{<span class="i">$AtomID1</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
549 536 <span class="s">}</span>
|
|
|
550 537 <span class="i">$This</span>->{<span class="w">BondSymbols</span>}{<span class="i">$AtomID1</span>}{<span class="i">$AtomID2</span>} = <span class="i">$BondSymbol</span><span class="sc">;</span>
|
|
|
551 538 <span class="s">}</span>
|
|
|
552 539 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
553 540 <span class="s">}</span>
|
|
|
554 541
|
|
|
555 542 <span class="c"># Get appropriate atom paths with length up to MaxLength...</span>
|
|
|
556 543 <span class="c">#</span>
|
|
|
557 <a name="_GenerateAtomPathsUpToMaxLength-"></a> 544 <span class="k">sub </span><span class="m">_GenerateAtomPathsUpToMaxLength</span> <span class="s">{</span>
|
|
|
558 545 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
559 546 <span class="k">my</span><span class="s">(</span><span class="i">$PathLength</span><span class="cm">,</span> <span class="i">$AllowRings</span><span class="cm">,</span> <span class="i">$Molecule</span><span class="cm">,</span> <span class="i">$AtomPathsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
560 547
|
|
|
561 548 <span class="i">$PathLength</span> = <span class="i">$This</span>->{<span class="w">MaxLength</span>}<span class="sc">;</span>
|
|
|
562 549 <span class="i">$AllowRings</span> = <span class="i">$This</span>->{<span class="w">AllowRings</span>}<span class="sc">;</span>
|
|
|
563 550 <span class="i">$Molecule</span> = <span class="i">$This</span>->{<span class="w">Molecule</span>}<span class="sc">;</span>
|
|
|
564 551
|
|
|
565 552 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AllowSharedBonds</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
566 553 <span class="i">$AtomPathsRef</span> = <span class="i">$Molecule</span><span class="i">->GetAllAtomPathsWithLengthUpto</span><span class="s">(</span><span class="i">$PathLength</span><span class="cm">,</span> <span class="i">$AllowRings</span><span class="s">)</span><span class="sc">;</span>
|
|
|
567 554 <span class="s">}</span>
|
|
|
568 555 <span class="k">else</span> <span class="s">{</span>
|
|
|
569 556 <span class="i">$AtomPathsRef</span> = <span class="i">$Molecule</span><span class="i">->GetAtomPathsWithLengthUpto</span><span class="s">(</span><span class="i">$PathLength</span><span class="cm">,</span> <span class="i">$AllowRings</span><span class="s">)</span><span class="sc">;</span>
|
|
|
570 557 <span class="s">}</span>
|
|
|
571 558 <span class="i">$This</span>->{<span class="w">AtomPathsRef</span>} = <span class="i">$AtomPathsRef</span><span class="sc">;</span>
|
|
|
572 559
|
|
|
573 560 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
574 561 <span class="s">}</span>
|
|
|
575 562
|
|
|
576 563 <span class="c"># Initialize atom paths strings at various pathlength levels...</span>
|
|
|
577 564 <span class="c">#</span>
|
|
|
578 <a name="_InitializeAtomPathsStrings-"></a> 565 <span class="k">sub </span><span class="m">_InitializeAtomPathsStrings</span> <span class="s">{</span>
|
|
|
579 566 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
580 567 <span class="k">my</span><span class="s">(</span><span class="i">$PathLength</span><span class="s">)</span><span class="sc">;</span>
|
|
|
581 568
|
|
|
582 569 <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
583 570
|
|
|
584 571 <span class="k">for</span> <span class="i">$PathLength</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">MinLength</span>} .. <span class="i">$This</span>->{<span class="w">MaxLength</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
585 572 <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
586 573 <span class="s">}</span>
|
|
|
587 574
|
|
|
588 575 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
589 576 <span class="s">}</span>
|
|
|
590 577
|
|
|
591 578 <span class="c"># Generate appropriate atom path strings for unique atom paths...</span>
|
|
|
592 579 <span class="c">#</span>
|
|
|
593 <a name="_GenerateAtomPathsStrings-"></a> 580 <span class="k">sub </span><span class="m">_GenerateAtomPathsStrings</span> <span class="s">{</span>
|
|
|
594 581 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
595 582 <span class="k">my</span><span class="s">(</span><span class="i">$PathLength</span><span class="cm">,</span> <span class="i">$MinPathLength</span><span class="cm">,</span> <span class="i">$UseUniquePaths</span><span class="s">)</span><span class="sc">;</span>
|
|
|
596 583
|
|
|
597 584 <span class="i">$MinPathLength</span> = <span class="i">$This</span>->{<span class="w">MinLength</span>}<span class="sc">;</span>
|
|
|
598 585 <span class="i">$UseUniquePaths</span> = <span class="i">$This</span>->{<span class="w">UseUniquePaths</span>}<span class="sc">;</span>
|
|
|
599 586
|
|
|
600 587 <span class="j">PATHATOMS:</span> <span class="k">for</span> <span class="i">$PathAtomsRef</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$This</span>->{<span class="w">AtomPathsRef</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
601 588 <span class="i">$PathLength</span> = <span class="k">scalar</span> <span class="i">@</span>{<span class="i">$PathAtomsRef</span>}<span class="sc">;</span>
|
|
|
602 589 <span class="k">if</span> <span class="s">(</span><span class="i">$PathLength</span> < <span class="i">$MinPathLength</span><span class="s">)</span> <span class="s">{</span>
|
|
|
603 590 <span class="k">next</span> <span class="j">PATHATOMS</span><span class="sc">;</span>
|
|
|
604 591 <span class="s">}</span>
|
|
|
605 592 <span class="k">if</span> <span class="s">(</span><span class="i">$UseUniquePaths</span><span class="s">)</span> <span class="s">{</span>
|
|
|
606 593 <span class="i">$This</span><span class="i">->_GenerateAtomPathStringUsingUniquePath</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
607 594 <span class="s">}</span>
|
|
|
608 595 <span class="k">else</span> <span class="s">{</span>
|
|
|
609 596 <span class="i">$This</span><span class="i">->_GenerateAtomPathString</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
610 597 <span class="s">}</span>
|
|
|
611 598 <span class="s">}</span>
|
|
|
612 599 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
613 600 <span class="s">}</span>
|
|
|
614 601
|
|
|
615 602 <span class="c"># Generate atom path string using unique path...</span>
|
|
|
616 603 <span class="c">#</span>
|
|
|
617 <a name="_GenerateAtomPathStringUsingUniquePath-"></a> 604 <span class="k">sub </span><span class="m">_GenerateAtomPathStringUsingUniquePath</span> <span class="s">{</span>
|
|
|
618 605 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
619 606
|
|
|
620 607 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AllowRings</span>} && <span class="i">$This</span><span class="i">->_DoesAtomPathContainsCycle</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
621 608 <span class="i">$This</span><span class="i">->_GenerateAtomPathStringUsingUniquePathContainingCycle</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
622 609 <span class="s">}</span>
|
|
|
623 610 <span class="k">else</span> <span class="s">{</span>
|
|
|
624 611 <span class="i">$This</span><span class="i">->_GenerateAtomPathStringUsingUniqueLinearPath</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
625 612 <span class="s">}</span>
|
|
|
626 613 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
627 614 <span class="s">}</span>
|
|
|
628 615
|
|
|
629 616 <span class="c"># Generate atom path string for specified path containing no cycle...</span>
|
|
|
630 617 <span class="c">#</span>
|
|
|
631 <a name="_GenerateAtomPathStringUsingUniqueLinearPath-"></a> 618 <span class="k">sub </span><span class="m">_GenerateAtomPathStringUsingUniqueLinearPath</span> <span class="s">{</span>
|
|
|
632 619 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
633 620
|
|
|
634 621 <span class="c"># Is it a unique linear atom path?</span>
|
|
|
635 622 <span class="c">#</span>
|
|
|
636 623 <span class="k">if</span> <span class="s">(</span>!<span class="i">$This</span><span class="i">->_IsUniqueLinearAtomPath</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
637 624 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
638 625 <span class="s">}</span>
|
|
|
639 626 <span class="i">$This</span><span class="i">->_GenerateAtomPathString</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
640 627
|
|
|
641 628 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
642 629 <span class="s">}</span>
|
|
|
643 630
|
|
|
644 631 <span class="c"># Is it a structurally unique linear path?</span>
|
|
|
645 632 <span class="c">#</span>
|
|
|
646 633 <span class="c"># For a path to be structurally unique, all of its atom IDs must be diffferent from any</span>
|
|
|
647 634 <span class="c"># earlier path atom IDs. In order to generate atom path atom ID invariant of the atom</span>
|
|
|
648 635 <span class="c"># order in the molecule, atom IDs are sorted numerically before generating the path ID.</span>
|
|
|
649 636 <span class="c">#</span>
|
|
|
650 637 <span class="c"># Notes:</span>
|
|
|
651 638 <span class="c"># . Atom path ID doesn't reflect the order of atoms in the atom path.</span>
|
|
|
652 639 <span class="c">#</span>
|
|
|
653 <a name="_IsUniqueLinearAtomPath-"></a> 640 <span class="k">sub </span><span class="m">_IsUniqueLinearAtomPath</span> <span class="s">{</span>
|
|
|
654 641 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
655 642 <span class="k">my</span><span class="s">(</span><span class="i">$AtomPathID</span><span class="cm">,</span> <span class="i">$PathLength</span><span class="cm">,</span> <span class="i">@PathAtomIDs</span><span class="s">)</span><span class="sc">;</span>
|
|
|
656 643
|
|
|
657 644 <span class="i">@PathAtomIDs</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
658 645 <span class="i">@PathAtomIDs</span> = <span class="k">map</span> <span class="s">{</span> <span class="i">$_</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> <span class="s">}</span> <span class="i">@</span>{<span class="i">$PathAtomsRef</span>}<span class="sc">;</span>
|
|
|
659 646
|
|
|
660 647 <span class="i">$AtomPathID</span> = <span class="k">join</span> <span class="q">'-'</span><span class="cm">,</span> <span class="k">sort</span> <span class="s">{</span> <span class="i">$a</span> <=> <span class="i">$b</span> <span class="s">}</span> <span class="i">@PathAtomIDs</span><span class="sc">;</span>
|
|
|
661 648 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$This</span>->{<span class="w">UniqueLinearAtomPathsIDs</span>}{<span class="i">$AtomPathID</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
662 649 <span class="k">return</span> <span class="n">0</span><span class="sc">;</span>
|
|
|
663 650 <span class="s">}</span>
|
|
|
664 651
|
|
|
665 652 <span class="c"># It's a unique atom path...</span>
|
|
|
666 653 <span class="i">$This</span>->{<span class="w">UniqueLinearAtomPathsIDs</span>}{<span class="i">$AtomPathID</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
667 654
|
|
|
668 655 <span class="k">return</span> <span class="n">1</span><span class="sc">;</span>
|
|
|
669 656 <span class="s">}</span>
|
|
|
670 657
|
|
|
671 658 <span class="c"># Generate atom path string for specified path containing a cycle...</span>
|
|
|
672 659 <span class="c">#</span>
|
|
|
673 <a name="_GenerateAtomPathStringUsingUniquePathContainingCycle-"></a> 660 <span class="k">sub </span><span class="m">_GenerateAtomPathStringUsingUniquePathContainingCycle</span> <span class="s">{</span>
|
|
|
674 661 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
675 662
|
|
|
676 663 <span class="c"># Is it a unique atom path containing a cycle?</span>
|
|
|
677 664 <span class="c">#</span>
|
|
|
678 665 <span class="k">if</span> <span class="s">(</span>!<span class="i">$This</span><span class="i">->_IsUniqueAtomPathContainingCycle</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
679 666 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
680 667 <span class="s">}</span>
|
|
|
681 668
|
|
|
682 669 <span class="k">my</span><span class="s">(</span><span class="i">$CycleClosingPathAtomIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
|
683 670 <span class="s">(</span><span class="i">$CycleClosingPathAtomIndex</span><span class="s">)</span> = <span class="i">$This</span><span class="i">->_GetAtomPathCycleClosingAtomIndex</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
684 671
|
|
|
685 672 <span class="k">if</span> <span class="s">(</span><span class="i">$CycleClosingPathAtomIndex</span> == <span class="n">0</span><span class="s">)</span> <span class="s">{</span>
|
|
|
686 673 <span class="i">$This</span><span class="i">->_GenerateUniqueAtomPathStringForPathCycle</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
687 674 <span class="s">}</span>
|
|
|
688 675 <span class="k">else</span> <span class="s">{</span>
|
|
|
689 676 <span class="i">$This</span><span class="i">->_GenerateUniqueAtomPathStringForPathContainingCycle</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="cm">,</span> <span class="i">$CycleClosingPathAtomIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
|
690 677 <span class="s">}</span>
|
|
|
691 678 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
692 679 <span class="s">}</span>
|
|
|
693 680
|
|
|
694 681 <span class="c"># Generate a unique atom path string for a cyclic path by generating atom path</span>
|
|
|
695 682 <span class="c"># strings for all possible paths in the cycle and keeping the lexicographically smallest</span>
|
|
|
696 683 <span class="c"># one.</span>
|
|
|
697 684 <span class="c">#</span>
|
|
|
698 685 <span class="c"># Although all the paths enumerated during atom path string generation are also</span>
|
|
|
699 686 <span class="c"># present in the intial paths list, but structural uniqueness check would detect</span>
|
|
|
700 687 <span class="c"># 'em earlier and this method ends being invoked only once for the first cyclic path.</span>
|
|
|
701 688 <span class="c">#</span>
|
|
|
702 689 <span class="c"># For atom paths containg same atom types and bond symbols, atom path strings</span>
|
|
|
703 690 <span class="c"># would be same for the paths.</span>
|
|
|
704 691 <span class="c">#</span>
|
|
|
705 <a name="_GenerateUniqueAtomPathStringForPathCycle-"></a> 692 <span class="k">sub </span><span class="m">_GenerateUniqueAtomPathStringForPathCycle</span> <span class="s">{</span>
|
|
|
706 693 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
707 694
|
|
|
708 695 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span><span class="i">->_AreAllPathAtomsSymbolsSame</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span> && <span class="i">$This</span><span class="i">->_AreAllPathBondSymbolsSame</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
709 696 <span class="k">return</span> <span class="i">$This</span><span class="i">->_GenerateAtomPathString</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
710 697 <span class="s">}</span>
|
|
|
711 698
|
|
|
712 699 <span class="c"># Generate all possible atom path strings and select the lexicographically smallest one...</span>
|
|
|
713 700 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$PathLength</span><span class="cm">,</span> <span class="i">$FinalAtomPathString</span><span class="cm">,</span> <span class="i">$FirstAtomPathString</span><span class="cm">,</span> <span class="i">$LastIndex</span><span class="cm">,</span> <span class="i">$FirstPartIndex</span><span class="cm">,</span> <span class="i">$FirstPartStartIndex</span><span class="cm">,</span> <span class="i">$FirstPartEndIndex</span><span class="cm">,</span> <span class="i">$SecondPartIndex</span><span class="cm">,</span> <span class="i">$SecondPartStartIndex</span><span class="cm">,</span> <span class="i">$SecondPartEndIndex</span><span class="cm">,</span> <span class="i">$AtomPathSymbolsRef</span><span class="cm">,</span> <span class="i">$AtomPathString</span><span class="cm">,</span> <span class="i">$ReverseAtomPathString</span><span class="cm">,</span> <span class="i">@FirstPartPathAtoms</span><span class="cm">,</span> <span class="i">@SecondPartPathAtoms</span><span class="cm">,</span> <span class="i">@PathAtoms</span><span class="s">)</span><span class="sc">;</span>
|
|
|
714 701
|
|
|
715 702 <span class="i">$PathLength</span> = <span class="k">scalar</span> <span class="i">@</span>{<span class="i">$PathAtomsRef</span>}<span class="sc">;</span>
|
|
|
716 703 <span class="i">$LastIndex</span> = <span class="i">$PathLength</span> - <span class="n">1</span><span class="sc">;</span>
|
|
|
717 704
|
|
|
718 705 <span class="i">$FinalAtomPathString</span> = <span class="q">''</span><span class="sc">;</span>
|
|
|
719 706 <span class="i">$FirstAtomPathString</span> = <span class="n">1</span><span class="sc">;</span>
|
|
|
720 707
|
|
|
721 708 <span class="i">@FirstPartPathAtoms</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> <span class="i">@SecondPartPathAtoms</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> <span class="i">@PathAtoms</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
722 709
|
|
|
723 710 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$LastIndex</span> - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
724 711 <span class="i">@FirstPartPathAtoms</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> <span class="i">@SecondPartPathAtoms</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> <span class="i">@PathAtoms</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
725 712
|
|
|
726 713 <span class="i">$FirstPartStartIndex</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$FirstPartEndIndex</span> = <span class="i">$Index</span> - <span class="n">1</span><span class="sc">;</span>
|
|
|
727 714 <span class="i">$SecondPartStartIndex</span> = <span class="i">$Index</span><span class="sc">;</span> <span class="i">$SecondPartEndIndex</span> = <span class="i">$LastIndex</span> - <span class="n">1</span><span class="sc">;</span>
|
|
|
728 715
|
|
|
729 716 <span class="c"># Get first part atoms...</span>
|
|
|
730 717 <span class="k">for</span> <span class="i">$FirstPartIndex</span> <span class="s">(</span><span class="i">$FirstPartStartIndex</span> .. <span class="i">$FirstPartEndIndex</span><span class="s">)</span> <span class="s">{</span>
|
|
|
731 718 <span class="k">push</span> <span class="i">@FirstPartPathAtoms</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span>->[<span class="i">$FirstPartIndex</span>]<span class="sc">;</span>
|
|
|
732 719 <span class="s">}</span>
|
|
|
733 720
|
|
|
734 721 <span class="c"># Get second part atoms...</span>
|
|
|
735 722 <span class="k">for</span> <span class="i">$SecondPartIndex</span> <span class="s">(</span><span class="i">$SecondPartStartIndex</span> .. <span class="i">$SecondPartEndIndex</span><span class="s">)</span> <span class="s">{</span>
|
|
|
736 723 <span class="k">push</span> <span class="i">@SecondPartPathAtoms</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span>->[<span class="i">$SecondPartIndex</span>]<span class="sc">;</span>
|
|
|
737 724 <span class="s">}</span>
|
|
|
738 725
|
|
|
739 726 <span class="c"># Get final list of path atoms...</span>
|
|
|
740 727 <span class="k">if</span> <span class="s">(</span><span class="i">@SecondPartPathAtoms</span><span class="s">)</span> <span class="s">{</span>
|
|
|
741 728 <span class="k">push</span> <span class="i">@PathAtoms</span><span class="cm">,</span> <span class="i">@SecondPartPathAtoms</span><span class="sc">;</span>
|
|
|
742 729 <span class="s">}</span>
|
|
|
743 730 <span class="k">if</span> <span class="s">(</span><span class="i">@FirstPartPathAtoms</span><span class="s">)</span> <span class="s">{</span>
|
|
|
744 731 <span class="k">push</span> <span class="i">@PathAtoms</span><span class="cm">,</span> <span class="i">@FirstPartPathAtoms</span><span class="sc">;</span>
|
|
|
745 732 <span class="s">}</span>
|
|
|
746 733
|
|
|
747 734 <span class="c"># Complete the cycle by adding first atom as the last atom...</span>
|
|
|
748 735 <span class="k">push</span> <span class="i">@PathAtoms</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span>->[<span class="i">$SecondPartStartIndex</span>]<span class="sc">;</span>
|
|
|
749 736
|
|
|
750 737 <span class="c"># Generate atom path string...</span>
|
|
|
751 738 <span class="i">$AtomPathSymbolsRef</span> = <span class="i">$This</span><span class="i">->_GenerateAtomPathSymbols</span><span class="s">(</span>\<span class="i">@PathAtoms</span><span class="s">)</span><span class="sc">;</span>
|
|
|
752 739
|
|
|
753 740 <span class="i">$AtomPathString</span> = <span class="k">join</span> <span class="q">''</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$AtomPathSymbolsRef</span>}<span class="sc">;</span>
|
|
|
754 741 <span class="i">$ReverseAtomPathString</span> = <span class="k">join</span> <span class="q">''</span><span class="cm">,</span> <span class="k">reverse</span> <span class="i">@</span>{<span class="i">$AtomPathSymbolsRef</span>}<span class="sc">;</span>
|
|
|
755 742
|
|
|
756 743 <span class="k">if</span> <span class="s">(</span><span class="i">$ReverseAtomPathString</span> <span class="k">le</span> <span class="i">$AtomPathString</span><span class="s">)</span> <span class="s">{</span>
|
|
|
757 744 <span class="i">$AtomPathString</span> = <span class="i">$ReverseAtomPathString</span><span class="sc">;</span>
|
|
|
758 745 <span class="s">}</span>
|
|
|
759 746
|
|
|
760 747 <span class="c"># Update final atom path string...</span>
|
|
|
761 748
|
|
|
762 749 <span class="k">if</span> <span class="s">(</span><span class="i">$FirstAtomPathString</span><span class="s">)</span> <span class="s">{</span>
|
|
|
763 750 <span class="i">$FirstAtomPathString</span> = <span class="n">0</span><span class="sc">;</span>
|
|
|
764 751 <span class="i">$FinalAtomPathString</span> = <span class="i">$AtomPathString</span><span class="sc">;</span>
|
|
|
765 752 <span class="s">}</span>
|
|
|
766 753 <span class="k">else</span> <span class="s">{</span>
|
|
|
767 754 <span class="k">if</span> <span class="s">(</span><span class="i">$AtomPathString</span> <span class="k">le</span> <span class="i">$FinalAtomPathString</span><span class="s">)</span> <span class="s">{</span>
|
|
|
768 755 <span class="i">$FinalAtomPathString</span> = <span class="i">$AtomPathString</span><span class="sc">;</span>
|
|
|
769 756 <span class="s">}</span>
|
|
|
770 757 <span class="s">}</span>
|
|
|
771 758 <span class="s">}</span>
|
|
|
772 759
|
|
|
773 760 <span class="c"># Set final atom path string...</span>
|
|
|
774 761 <span class="c">#</span>
|
|
|
775 762 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$FinalAtomPathString</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
776 763 <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$FinalAtomPathString</span>} += <span class="n">1</span><span class="sc">;</span>
|
|
|
777 764 <span class="s">}</span>
|
|
|
778 765 <span class="k">else</span> <span class="s">{</span>
|
|
|
779 766 <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$FinalAtomPathString</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
780 767 <span class="s">}</span>
|
|
|
781 768
|
|
|
782 769 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
783 770 <span class="s">}</span>
|
|
|
784 771
|
|
|
785 772 <span class="c">#</span>
|
|
|
786 773 <span class="c"># Generate a unique atom path string for paths containing a cycle closed by</span>
|
|
|
787 774 <span class="c"># the specified atom index and the last atom index.</span>
|
|
|
788 775 <span class="c">#</span>
|
|
|
789 776 <span class="c"># The following methodology is used to generate atom path string which is</span>
|
|
|
790 777 <span class="c"># independemt of initial atom ordering:</span>
|
|
|
791 778 <span class="c"># . Generate atom paths string from first atom to the atom before the first cycle</span>
|
|
|
792 779 <span class="c"># closing atom.</span>
|
|
|
793 780 <span class="c"># . Generate atom path string from atoms from first cycle closing atom index to</span>
|
|
|
794 781 <span class="c"># the last path atom in both forward and reverse order. And select the lexicographically</span>
|
|
|
795 782 <span class="c"># smallest atom path string.</span>
|
|
|
796 783 <span class="c"># . Combine atom path string generated in first step with second step to generate</span>
|
|
|
797 784 <span class="c"># final atom path string.</span>
|
|
|
798 785 <span class="c">#</span>
|
|
|
799 <a name="_GenerateUniqueAtomPathStringForPathContainingCycle-"></a> 786 <span class="k">sub </span><span class="m">_GenerateUniqueAtomPathStringForPathContainingCycle</span> <span class="s">{</span>
|
|
|
800 787 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="cm">,</span> <span class="i">$CycleClosingAtomIndex</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
801 788 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$PathLength</span><span class="cm">,</span> <span class="i">$LastIndex</span><span class="cm">,</span> <span class="i">$LinearPartStartIndex</span><span class="cm">,</span> <span class="i">$LinearPartEndIndex</span><span class="cm">,</span> <span class="i">$CyclicPartStartIndex</span><span class="cm">,</span> <span class="i">$CyclicPartEndIndex</span><span class="cm">,</span> <span class="i">$CyclicPartAtomPathSymbolsRef</span><span class="cm">,</span> <span class="i">$CyclicPartAtomPathString</span><span class="cm">,</span> <span class="i">$ReverseCyclicPartAtomPathString</span><span class="cm">,</span> <span class="i">$AtomPathString</span><span class="cm">,</span> <span class="i">$AtomPathSymbolsRef</span><span class="cm">,</span> <span class="i">@CyclicPartPathAtoms</span><span class="cm">,</span> <span class="i">@PathAtoms</span><span class="s">)</span><span class="sc">;</span>
|
|
|
802 789
|
|
|
803 790 <span class="i">$PathLength</span> = <span class="k">scalar</span> <span class="i">@</span>{<span class="i">$PathAtomsRef</span>}<span class="sc">;</span>
|
|
|
804 791 <span class="i">$LastIndex</span> = <span class="i">$PathLength</span> - <span class="n">1</span><span class="sc">;</span>
|
|
|
805 792
|
|
|
806 793 <span class="i">@PathAtoms</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
807 794
|
|
|
808 795 <span class="c"># Get path atoms corresponding to linear part of the path...</span>
|
|
|
809 796 <span class="i">$LinearPartStartIndex</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$LinearPartEndIndex</span> = <span class="i">$CycleClosingAtomIndex</span> - <span class="n">1</span><span class="sc">;</span>
|
|
|
810 797
|
|
|
811 798 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="i">$LinearPartStartIndex</span> .. <span class="i">$LinearPartEndIndex</span><span class="s">)</span> <span class="s">{</span>
|
|
|
812 799 <span class="k">push</span> <span class="i">@PathAtoms</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span>->[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
|
813 800 <span class="s">}</span>
|
|
|
814 801
|
|
|
815 802 <span class="c"># Get atoms correcponding to cyclic part of the path...</span>
|
|
|
816 803 <span class="i">@CyclicPartPathAtoms</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
817 804 <span class="i">$CyclicPartStartIndex</span> = <span class="i">$CycleClosingAtomIndex</span><span class="sc">;</span> <span class="i">$CyclicPartEndIndex</span> = <span class="i">$LastIndex</span><span class="sc">;</span>
|
|
|
818 805
|
|
|
819 806 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="i">$CyclicPartStartIndex</span> .. <span class="i">$CyclicPartEndIndex</span><span class="s">)</span> <span class="s">{</span>
|
|
|
820 807 <span class="k">push</span> <span class="i">@CyclicPartPathAtoms</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span>->[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
|
821 808 <span class="s">}</span>
|
|
|
822 809
|
|
|
823 810 <span class="c"># Setup a lexicographically smaller atom path string for cyclic part...</span>
|
|
|
824 811
|
|
|
825 812 <span class="i">$CyclicPartAtomPathSymbolsRef</span> = <span class="i">$This</span><span class="i">->_GenerateAtomPathSymbols</span><span class="s">(</span>\<span class="i">@CyclicPartPathAtoms</span><span class="s">)</span><span class="sc">;</span>
|
|
|
826 813 <span class="i">$CyclicPartAtomPathString</span> = <span class="k">join</span> <span class="q">''</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$CyclicPartAtomPathSymbolsRef</span>}<span class="sc">;</span>
|
|
|
827 814 <span class="i">$ReverseCyclicPartAtomPathString</span> = <span class="k">join</span> <span class="q">''</span><span class="cm">,</span> <span class="k">reverse</span> <span class="i">@</span>{<span class="i">$CyclicPartAtomPathSymbolsRef</span>}<span class="sc">;</span>
|
|
|
828 815
|
|
|
829 816 <span class="c"># Setup atom path corresponding to linear part and lexigraphicall smaller cyclic part...</span>
|
|
|
830 817
|
|
|
831 818 <span class="k">if</span> <span class="s">(</span><span class="i">$ReverseCyclicPartAtomPathString</span> <span class="k">le</span> <span class="i">$CyclicPartAtomPathString</span><span class="s">)</span> <span class="s">{</span>
|
|
|
832 819 <span class="k">push</span> <span class="i">@PathAtoms</span><span class="cm">,</span> <span class="k">reverse</span> <span class="i">@CyclicPartPathAtoms</span><span class="sc">;</span>
|
|
|
833 820 <span class="s">}</span>
|
|
|
834 821 <span class="k">else</span> <span class="s">{</span>
|
|
|
835 822 <span class="k">push</span> <span class="i">@PathAtoms</span><span class="cm">,</span> <span class="i">@CyclicPartPathAtoms</span><span class="sc">;</span>
|
|
|
836 823 <span class="s">}</span>
|
|
|
837 824
|
|
|
838 825 <span class="c"># Setup final atom path string...</span>
|
|
|
839 826
|
|
|
840 827 <span class="i">$AtomPathSymbolsRef</span> = <span class="i">$This</span><span class="i">->_GenerateAtomPathSymbols</span><span class="s">(</span>\<span class="i">@PathAtoms</span><span class="s">)</span><span class="sc">;</span>
|
|
|
841 828 <span class="i">$AtomPathString</span> = <span class="k">join</span> <span class="q">''</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$AtomPathSymbolsRef</span>}<span class="sc">;</span>
|
|
|
842 829
|
|
|
843 830 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$AtomPathString</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
844 831 <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$AtomPathString</span>} += <span class="n">1</span><span class="sc">;</span>
|
|
|
845 832 <span class="s">}</span>
|
|
|
846 833 <span class="k">else</span> <span class="s">{</span>
|
|
|
847 834 <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$AtomPathString</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
848 835 <span class="s">}</span>
|
|
|
849 836
|
|
|
850 837 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
851 838 <span class="s">}</span>
|
|
|
852 839
|
|
|
853 840 <span class="c"># Does atom path contain a cycle?</span>
|
|
|
854 841 <span class="c">#</span>
|
|
|
855 842 <span class="c"># For an atom path to contain cycle, it must satisfy the following conditions:</span>
|
|
|
856 843 <span class="c"># . Pathlength >= 3</span>
|
|
|
857 844 <span class="c"># . Last atom ID is equal to first atom ID or some other atom ID besides itself</span>
|
|
|
858 845 <span class="c">#</span>
|
|
|
859 <a name="_DoesAtomPathContainsCycle-"></a> 846 <span class="k">sub </span><span class="m">_DoesAtomPathContainsCycle</span> <span class="s">{</span>
|
|
|
860 847 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
861 848 <span class="k">my</span><span class="s">(</span><span class="i">$PathLength</span><span class="s">)</span><span class="sc">;</span>
|
|
|
862 849
|
|
|
863 850 <span class="i">$PathLength</span> = <span class="k">scalar</span> <span class="i">@</span>{<span class="i">$PathAtomsRef</span>}<span class="sc">;</span>
|
|
|
864 851 <span class="k">if</span> <span class="s">(</span><span class="i">$PathLength</span> <= <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
|
865 852 <span class="k">return</span> <span class="n">0</span><span class="sc">;</span>
|
|
|
866 853 <span class="s">}</span>
|
|
|
867 854
|
|
|
868 855 <span class="k">my</span><span class="s">(</span><span class="i">$AtomIndex</span><span class="cm">,</span> <span class="i">$LastAtomIndex</span><span class="cm">,</span> <span class="i">$Atom</span><span class="cm">,</span> <span class="i">$AtomID</span><span class="cm">,</span> <span class="i">$LastAtom</span><span class="cm">,</span> <span class="i">$LastAtomID</span><span class="s">)</span><span class="sc">;</span>
|
|
|
869 856
|
|
|
870 857 <span class="i">$LastAtomIndex</span> = <span class="i">$PathLength</span> - <span class="n">1</span><span class="sc">;</span>
|
|
|
871 858 <span class="i">$LastAtom</span> = <span class="i">$PathAtomsRef</span>->[<span class="i">$LastAtomIndex</span>]<span class="sc">;</span>
|
|
|
872 859 <span class="i">$LastAtomID</span> = <span class="i">$LastAtom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
873 860
|
|
|
874 861 <span class="c"># Look for atomID similar to last atom ID...</span>
|
|
|
875 862 <span class="k">for</span> <span class="i">$AtomIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$LastAtomIndex</span> - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
876 863 <span class="i">$Atom</span> = <span class="i">$PathAtomsRef</span>->[<span class="i">$AtomIndex</span>]<span class="sc">;</span>
|
|
|
877 864 <span class="i">$AtomID</span> = <span class="i">$Atom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
878 865
|
|
|
879 866 <span class="k">if</span> <span class="s">(</span><span class="i">$AtomID</span> == <span class="i">$LastAtomID</span><span class="s">)</span> <span class="s">{</span>
|
|
|
880 867 <span class="c"># It's a cycle...</span>
|
|
|
881 868 <span class="k">return</span> <span class="n">1</span><span class="sc">;</span>
|
|
|
882 869 <span class="s">}</span>
|
|
|
883 870 <span class="s">}</span>
|
|
|
884 871 <span class="k">return</span> <span class="n">0</span><span class="sc">;</span>
|
|
|
885 872 <span class="s">}</span>
|
|
|
886 873
|
|
|
887 874 <span class="c"># Get atom path cycle closing atom index...</span>
|
|
|
888 875 <span class="c">#</span>
|
|
|
889 <a name="_GetAtomPathCycleClosingAtomIndex-"></a> 876 <span class="k">sub </span><span class="m">_GetAtomPathCycleClosingAtomIndex</span> <span class="s">{</span>
|
|
|
890 877 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
891 878 <span class="k">my</span><span class="s">(</span><span class="i">$AtomIndex</span><span class="cm">,</span> <span class="i">$LastAtomIndex</span><span class="cm">,</span> <span class="i">$Atom</span><span class="cm">,</span> <span class="i">$AtomID</span><span class="cm">,</span> <span class="i">$LastAtom</span><span class="cm">,</span> <span class="i">$LastAtomID</span><span class="cm">,</span> <span class="i">$PathLength</span><span class="s">)</span><span class="sc">;</span>
|
|
|
892 879
|
|
|
893 880 <span class="i">$PathLength</span> = <span class="k">scalar</span> <span class="i">@</span>{<span class="i">$PathAtomsRef</span>}<span class="sc">;</span>
|
|
|
894 881
|
|
|
895 882 <span class="i">$LastAtomIndex</span> = <span class="i">$PathLength</span> - <span class="n">1</span><span class="sc">;</span>
|
|
|
896 883 <span class="i">$LastAtom</span> = <span class="i">$PathAtomsRef</span>->[<span class="i">$LastAtomIndex</span>]<span class="sc">;</span> <span class="i">$LastAtomID</span> = <span class="i">$LastAtom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
897 884
|
|
|
898 885 <span class="c"># Look for atomID similar to last atom ID...</span>
|
|
|
899 886 <span class="k">for</span> <span class="i">$AtomIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$LastAtomIndex</span> - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
900 887 <span class="i">$Atom</span> = <span class="i">$PathAtomsRef</span>->[<span class="i">$AtomIndex</span>]<span class="sc">;</span> <span class="i">$AtomID</span> = <span class="i">$Atom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
901 888
|
|
|
902 889 <span class="k">if</span> <span class="s">(</span><span class="i">$AtomID</span> == <span class="i">$LastAtomID</span><span class="s">)</span> <span class="s">{</span>
|
|
|
903 890 <span class="c"># It's a cycle closing atom...</span>
|
|
|
904 891 <span class="k">return</span> <span class="i">$AtomIndex</span><span class="sc">;</span>
|
|
|
905 892 <span class="s">}</span>
|
|
|
906 893 <span class="s">}</span>
|
|
|
907 894 <span class="k">return</span> <span class="k">undef</span><span class="sc">;</span>
|
|
|
908 895 <span class="s">}</span>
|
|
|
909 896
|
|
|
910 897 <span class="c"># Is it a structurally unique path containing a cycle?</span>
|
|
|
911 898 <span class="c">#</span>
|
|
|
912 899 <span class="c"># For atom paths containing cycles, last atom ID is either equal to first atom ID or</span>
|
|
|
913 900 <span class="c"># some other atom ID besides itself.</span>
|
|
|
914 901 <span class="c">#</span>
|
|
|
915 902 <span class="c"># In order to determine its structurally unqiue independent of initial atom ordering,</span>
|
|
|
916 903 <span class="c"># the following methodolgy is used:</span>
|
|
|
917 904 <span class="c">#</span>
|
|
|
918 905 <span class="c"># . For paths with same first and atom IDs:</span>
|
|
|
919 906 <span class="c"># . Remove the last atom ID from atom path</span>
|
|
|
920 907 <span class="c"># . Sort atom IDs in the path</span>
|
|
|
921 908 <span class="c"># . Add first atom ID from the sorted list to the end of list to complete the cycle</span>
|
|
|
922 909 <span class="c"># . Generate a atom path ID</span>
|
|
|
923 910 <span class="c"># . Use final path ID to track uniqueness of path containing cycle.</span>
|
|
|
924 911 <span class="c">#</span>
|
|
|
925 912 <span class="c"># . For paths with last atom ID equal to some other atom ID besidies itself:</span>
|
|
|
926 913 <span class="c"># . Sort atom IDs in atom path</span>
|
|
|
927 914 <span class="c"># . Generate atom path ID and use it to track unqiueness of atom paths.</span>
|
|
|
928 915 <span class="c">#</span>
|
|
|
929 <a name="_IsUniqueAtomPathContainingCycle-"></a> 916 <span class="k">sub </span><span class="m">_IsUniqueAtomPathContainingCycle</span> <span class="s">{</span>
|
|
|
930 917 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
931 918 <span class="k">my</span><span class="s">(</span><span class="i">$PathLength</span><span class="cm">,</span> <span class="i">$AtomPathID</span><span class="cm">,</span> <span class="i">$FirstAtom</span><span class="cm">,</span> <span class="i">$LastAtom</span><span class="cm">,</span> <span class="i">$FirstAtomID</span><span class="cm">,</span> <span class="i">$LastAtomID</span><span class="cm">,</span> <span class="i">@PathAtomIDs</span><span class="cm">,</span> <span class="i">@SortedPathAtomIDs</span><span class="s">)</span><span class="sc">;</span>
|
|
|
932 919
|
|
|
933 920 <span class="i">@PathAtomIDs</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
934 921 <span class="i">@PathAtomIDs</span> = <span class="k">map</span> <span class="s">{</span> <span class="i">$_</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> <span class="s">}</span> <span class="i">@</span>{<span class="i">$PathAtomsRef</span>}<span class="sc">;</span>
|
|
|
935 922
|
|
|
936 923 <span class="i">$PathLength</span> = <span class="k">scalar</span> <span class="i">@</span>{<span class="i">$PathAtomsRef</span>}<span class="sc">;</span>
|
|
|
937 924
|
|
|
938 925 <span class="i">$FirstAtom</span> = <span class="i">$PathAtomsRef</span>->[<span class="n">0</span>]<span class="sc">;</span> <span class="i">$FirstAtomID</span> = <span class="i">$FirstAtom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
939 926 <span class="i">$LastAtom</span> = <span class="i">$PathAtomsRef</span>->[<span class="i">$PathLength</span> - <span class="n">1</span>]<span class="sc">;</span> <span class="i">$LastAtomID</span> = <span class="i">$LastAtom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
940 927
|
|
|
941 928 <span class="k">if</span> <span class="s">(</span><span class="i">$FirstAtomID</span> == <span class="i">$LastAtomID</span><span class="s">)</span> <span class="s">{</span>
|
|
|
942 929 <span class="k">pop</span> <span class="i">@PathAtomIDs</span><span class="sc">;</span>
|
|
|
943 930
|
|
|
944 931 <span class="i">@SortedPathAtomIDs</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
945 932 <span class="i">@SortedPathAtomIDs</span> = <span class="k">sort</span> <span class="s">{</span> <span class="i">$a</span> <=> <span class="i">$b</span> <span class="s">}</span> <span class="i">@PathAtomIDs</span><span class="sc">;</span>
|
|
|
946 933
|
|
|
947 934 <span class="k">push</span> <span class="i">@SortedPathAtomIDs</span><span class="cm">,</span> <span class="i">$SortedPathAtomIDs</span>[<span class="n">0</span>]<span class="sc">;</span>
|
|
|
948 935
|
|
|
949 936 <span class="i">$AtomPathID</span> = <span class="k">join</span> <span class="q">'-'</span><span class="cm">,</span> <span class="i">@SortedPathAtomIDs</span><span class="sc">;</span>
|
|
|
950 937 <span class="s">}</span>
|
|
|
951 938 <span class="k">else</span> <span class="s">{</span>
|
|
|
952 939 <span class="i">$AtomPathID</span> = <span class="k">join</span> <span class="q">'-'</span><span class="cm">,</span> <span class="k">sort</span> <span class="s">{</span> <span class="i">$a</span> <=> <span class="i">$b</span> <span class="s">}</span> <span class="i">@PathAtomIDs</span><span class="sc">;</span>
|
|
|
953 940 <span class="s">}</span>
|
|
|
954 941
|
|
|
955 942 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$This</span>->{<span class="w">UniqueCyclicAtomPathsIDs</span>}{<span class="i">$AtomPathID</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
956 943 <span class="k">return</span> <span class="n">0</span><span class="sc">;</span>
|
|
|
957 944 <span class="s">}</span>
|
|
|
958 945
|
|
|
959 946 <span class="c"># It's a unique atom path containing a cycle...</span>
|
|
|
960 947 <span class="i">$This</span>->{<span class="w">UniqueCyclicAtomPathsIDs</span>}{<span class="i">$AtomPathID</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
961 948
|
|
|
962 949 <span class="k">return</span> <span class="n">1</span><span class="sc">;</span>
|
|
|
963 950 <span class="s">}</span>
|
|
|
964 951
|
|
|
965 952 <span class="c"># Generate atom path string for specified atom path...</span>
|
|
|
966 953 <span class="c">#</span>
|
|
|
967 <a name="_GenerateAtomPathString-"></a> 954 <span class="k">sub </span><span class="m">_GenerateAtomPathString</span> <span class="s">{</span>
|
|
|
968 955 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
969 956 <span class="k">my</span><span class="s">(</span><span class="i">$PathLength</span><span class="cm">,</span> <span class="i">$AtomPathString</span><span class="cm">,</span> <span class="i">$ReverseAtomPathString</span><span class="cm">,</span> <span class="i">$AtomPathSymbolsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
970 957
|
|
|
971 958 <span class="i">$PathLength</span> = <span class="k">scalar</span> <span class="i">@</span>{<span class="i">$PathAtomsRef</span>}<span class="sc">;</span>
|
|
|
972 959
|
|
|
973 960 <span class="c"># Generate path atom and bond symbols...</span>
|
|
|
974 961 <span class="c">#</span>
|
|
|
975 962 <span class="i">$AtomPathSymbolsRef</span> = <span class="i">$This</span><span class="i">->_GenerateAtomPathSymbols</span><span class="s">(</span><span class="i">$PathAtomsRef</span><span class="s">)</span><span class="sc">;</span>
|
|
|
976 963
|
|
|
977 964 <span class="c"># Check presence of path using path ID created by atom path symbols...</span>
|
|
|
978 965 <span class="i">$AtomPathString</span> = <span class="k">join</span> <span class="q">''</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$AtomPathSymbolsRef</span>}<span class="sc">;</span>
|
|
|
979 966 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$AtomPathString</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
980 967 <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$AtomPathString</span>} += <span class="n">1</span><span class="sc">;</span>
|
|
|
981 968 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
982 969 <span class="s">}</span>
|
|
|
983 970
|
|
|
984 971 <span class="c"># Check presence of reverse path using path ID created by atom path symbols...</span>
|
|
|
985 972 <span class="c">#</span>
|
|
|
986 973 <span class="i">$ReverseAtomPathString</span> = <span class="k">join</span> <span class="q">''</span><span class="cm">,</span> <span class="k">reverse</span> <span class="i">@</span>{<span class="i">$AtomPathSymbolsRef</span>}<span class="sc">;</span>
|
|
|
987 974 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$ReverseAtomPathString</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
988 975 <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$ReverseAtomPathString</span>} += <span class="n">1</span><span class="sc">;</span>
|
|
|
989 976 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
990 977 <span class="s">}</span>
|
|
|
991 978
|
|
|
992 979 <span class="c"># Use lexicographically smaller atom path string as PathID...</span>
|
|
|
993 980 <span class="c">#</span>
|
|
|
994 981 <span class="k">if</span> <span class="s">(</span><span class="i">$AtomPathString</span> <span class="k">le</span> <span class="i">$ReverseAtomPathString</span><span class="s">)</span> <span class="s">{</span>
|
|
|
995 982 <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$AtomPathString</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
996 983 <span class="s">}</span>
|
|
|
997 984 <span class="k">else</span> <span class="s">{</span>
|
|
|
998 985 <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$ReverseAtomPathString</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
999 986 <span class="s">}</span>
|
|
|
1000 987 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
1001 988 <span class="s">}</span>
|
|
|
1002 989
|
|
|
1003 990 <span class="c"># Are atom types for all path atoms same?</span>
|
|
|
1004 991 <span class="c">#</span>
|
|
|
1005 <a name="_AreAllPathAtomsSymbolsSame-"></a> 992 <span class="k">sub </span><span class="m">_AreAllPathAtomsSymbolsSame</span> <span class="s">{</span>
|
|
|
1006 993 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1007 994 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$Atom</span><span class="cm">,</span> <span class="i">$AtomID</span><span class="cm">,</span> <span class="i">$AtomType</span><span class="cm">,</span> <span class="i">$FirstAtomType</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1008 995
|
|
|
1009 996 <span class="i">$Atom</span> = <span class="i">$PathAtomsRef</span>->[<span class="n">0</span>]<span class="sc">;</span> <span class="i">$AtomID</span> = <span class="i">$Atom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1010 997 <span class="i">$FirstAtomType</span> = <span class="i">$This</span>->{<span class="w">AssignedAtomTypes</span>}{<span class="i">$AtomID</span>}<span class="sc">;</span>
|
|
|
1011 998
|
|
|
1012 999 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">1</span> .. <span class="i">$#</span>{<span class="i">$PathAtomsRef</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
1013 1000 <span class="i">$Atom</span> = <span class="i">$PathAtomsRef</span>->[<span class="i">$Index</span>]<span class="sc">;</span> <span class="i">$AtomID</span> = <span class="i">$Atom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1014 1001 <span class="i">$AtomType</span> = <span class="i">$This</span>->{<span class="w">AssignedAtomTypes</span>}{<span class="i">$AtomID</span>}<span class="sc">;</span>
|
|
|
1015 1002
|
|
|
1016 1003 <span class="k">if</span> <span class="s">(</span><span class="i">$AtomType</span> <span class="k">ne</span> <span class="i">$FirstAtomType</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1017 1004 <span class="k">return</span> <span class="n">0</span><span class="sc">;</span>
|
|
|
1018 1005 <span class="s">}</span>
|
|
|
1019 1006 <span class="s">}</span>
|
|
|
1020 1007 <span class="k">return</span> <span class="n">1</span><span class="sc">;</span>
|
|
|
1021 1008 <span class="s">}</span>
|
|
|
1022 1009
|
|
|
1023 1010 <span class="c"># Are bond symbols for all path bonds same?</span>
|
|
|
1024 1011 <span class="c">#</span>
|
|
|
1025 <a name="_AreAllPathBondSymbolsSame-"></a>1012 <span class="k">sub </span><span class="m">_AreAllPathBondSymbolsSame</span> <span class="s">{</span>
|
|
|
1026 1013 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1027 1014 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$Atom</span><span class="cm">,</span> <span class="i">$BondedAtom</span><span class="cm">,</span> <span class="i">$AtomID</span><span class="cm">,</span> <span class="i">$BondedAtomID</span><span class="cm">,</span> <span class="i">$BondAtomID1</span><span class="cm">,</span> <span class="i">$BondAtomID2</span><span class="cm">,</span> <span class="i">$FirstBondSymbol</span><span class="cm">,</span> <span class="i">$BondSymbol</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1028 1015
|
|
|
1029 1016 <span class="c"># During no usage of bond symbols, just ignore them and assume they are same...</span>
|
|
|
1030 1017 <span class="k">if</span> <span class="s">(</span>!<span class="i">$This</span>->{<span class="w">UseBondSymbols</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
1031 1018 <span class="k">return</span> <span class="n">1</span><span class="sc">;</span>
|
|
|
1032 1019 <span class="s">}</span>
|
|
|
1033 1020
|
|
|
1034 1021 <span class="i">$Atom</span> = <span class="i">$PathAtomsRef</span>->[<span class="n">0</span>]<span class="sc">;</span> <span class="i">$BondedAtom</span> = <span class="i">$PathAtomsRef</span>->[<span class="n">1</span>]<span class="sc">;</span>
|
|
|
1035 1022 <span class="i">$AtomID</span> = <span class="i">$Atom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> <span class="i">$BondedAtomID</span> = <span class="i">$BondedAtom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1036 1023
|
|
|
1037 1024 <span class="s">(</span><span class="i">$BondAtomID1</span><span class="cm">,</span> <span class="i">$BondAtomID2</span><span class="s">)</span> = <span class="s">(</span><span class="i">$AtomID</span> < <span class="i">$BondedAtomID</span><span class="s">)</span> ? <span class="s">(</span><span class="i">$AtomID</span><span class="cm">,</span> <span class="i">$BondedAtomID</span><span class="s">)</span> <span class="co">:</span> <span class="s">(</span><span class="i">$BondedAtomID</span><span class="cm">,</span> <span class="i">$AtomID</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1038 1025 <span class="i">$FirstBondSymbol</span> = <span class="i">$This</span>->{<span class="w">BondSymbols</span>}{<span class="i">$BondAtomID1</span>}{<span class="i">$BondAtomID2</span>}<span class="sc">;</span>
|
|
|
1039 1026
|
|
|
1040 1027 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">1</span> .. <span class="s">(</span><span class="i">$#</span>{<span class="i">$PathAtomsRef</span>} - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1041 1028 <span class="i">$Atom</span> = <span class="i">$PathAtomsRef</span>->[<span class="i">$Index</span>]<span class="sc">;</span> <span class="i">$BondedAtom</span> = <span class="i">$PathAtomsRef</span>->[<span class="i">$Index</span> + <span class="n">1</span>]<span class="sc">;</span>
|
|
|
1042 1029 <span class="i">$AtomID</span> = <span class="i">$Atom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> <span class="i">$BondedAtomID</span> = <span class="i">$BondedAtom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1043 1030
|
|
|
1044 1031 <span class="s">(</span><span class="i">$BondAtomID1</span><span class="cm">,</span> <span class="i">$BondAtomID2</span><span class="s">)</span> = <span class="s">(</span><span class="i">$AtomID</span> < <span class="i">$BondedAtomID</span><span class="s">)</span> ? <span class="s">(</span><span class="i">$AtomID</span><span class="cm">,</span> <span class="i">$BondedAtomID</span><span class="s">)</span> <span class="co">:</span> <span class="s">(</span><span class="i">$BondedAtomID</span><span class="cm">,</span> <span class="i">$AtomID</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1045 1032 <span class="i">$BondSymbol</span> = <span class="i">$This</span>->{<span class="w">BondSymbols</span>}{<span class="i">$BondAtomID1</span>}{<span class="i">$BondAtomID2</span>}<span class="sc">;</span>
|
|
|
1046 1033
|
|
|
1047 1034 <span class="k">if</span> <span class="s">(</span><span class="i">$BondSymbol</span> <span class="k">ne</span> <span class="i">$FirstBondSymbol</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1048 1035 <span class="k">return</span> <span class="n">0</span><span class="sc">;</span>
|
|
|
1049 1036 <span class="s">}</span>
|
|
|
1050 1037 <span class="s">}</span>
|
|
|
1051 1038 <span class="k">return</span> <span class="n">1</span><span class="sc">;</span>
|
|
|
1052 1039 <span class="s">}</span>
|
|
|
1053 1040
|
|
|
1054 1041 <span class="c"># Generate atom path symbols...</span>
|
|
|
1055 1042 <span class="c">#</span>
|
|
|
1056 <a name="_GenerateAtomPathSymbols-"></a>1043 <span class="k">sub </span><span class="m">_GenerateAtomPathSymbols</span> <span class="s">{</span>
|
|
|
1057 1044 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">$PathAtomsRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1058 1045 <span class="k">my</span><span class="s">(</span><span class="i">$Atom</span><span class="cm">,</span> <span class="i">$AtomID</span><span class="cm">,</span> <span class="i">@AtomPathSymbols</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1059 1046
|
|
|
1060 1047 <span class="i">@AtomPathSymbols</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1061 1048
|
|
|
1062 1049 <span class="k">if</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$PathAtomsRef</span>} == <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1063 1050 <span class="i">$Atom</span> = <span class="i">$PathAtomsRef</span>->[<span class="n">0</span>]<span class="sc">;</span> <span class="i">$AtomID</span> = <span class="i">$Atom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1064 1051 <span class="k">push</span> <span class="i">@AtomPathSymbols</span><span class="cm">,</span> <span class="i">$This</span>->{<span class="w">AssignedAtomTypes</span>}{<span class="i">$AtomID</span>}<span class="sc">;</span>
|
|
|
1065 1052 <span class="k">return</span> \<span class="i">@AtomPathSymbols</span><span class="sc">;</span>
|
|
|
1066 1053 <span class="s">}</span>
|
|
|
1067 1054
|
|
|
1068 1055 <span class="c"># Ignore bond information...</span>
|
|
|
1069 1056 <span class="k">if</span> <span class="s">(</span>!<span class="i">$This</span>->{<span class="w">UseBondSymbols</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
1070 1057 <span class="k">for</span> <span class="i">$Atom</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$PathAtomsRef</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
1071 1058 <span class="i">$AtomID</span> = <span class="i">$Atom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1072 1059 <span class="k">push</span> <span class="i">@AtomPathSymbols</span><span class="cm">,</span> <span class="i">$This</span>->{<span class="w">AssignedAtomTypes</span>}{<span class="i">$AtomID</span>}<span class="sc">;</span>
|
|
|
1073 1060 <span class="s">}</span>
|
|
|
1074 1061 <span class="k">return</span> \<span class="i">@AtomPathSymbols</span><span class="sc">;</span>
|
|
|
1075 1062 <span class="s">}</span>
|
|
|
1076 1063
|
|
|
1077 1064 <span class="c"># Use atoms and bonds to generate atom path string...</span>
|
|
|
1078 1065 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$BondedAtom</span><span class="cm">,</span> <span class="i">$BondedAtomID</span><span class="cm">,</span> <span class="i">$BondAtomID1</span><span class="cm">,</span> <span class="i">$BondAtomID2</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1079 1066
|
|
|
1080 1067 <span class="c"># Process atom type of first atom in path...</span>
|
|
|
1081 1068 <span class="i">$Atom</span> = <span class="i">$PathAtomsRef</span>->[<span class="n">0</span>]<span class="sc">;</span> <span class="i">$AtomID</span> = <span class="i">$Atom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1082 1069 <span class="k">push</span> <span class="i">@AtomPathSymbols</span><span class="cm">,</span> <span class="i">$This</span>->{<span class="w">AssignedAtomTypes</span>}{<span class="i">$AtomID</span>}<span class="sc">;</span>
|
|
|
1083 1070
|
|
|
1084 1071 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$#</span>{<span class="i">$PathAtomsRef</span>} - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1085 1072 <span class="i">$Atom</span> = <span class="i">$PathAtomsRef</span>->[<span class="i">$Index</span>]<span class="sc">;</span> <span class="i">$BondedAtom</span> = <span class="i">$PathAtomsRef</span>->[<span class="i">$Index</span> + <span class="n">1</span>]<span class="sc">;</span>
|
|
|
1086 1073 <span class="i">$AtomID</span> = <span class="i">$Atom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> <span class="i">$BondedAtomID</span> = <span class="i">$BondedAtom</span><span class="i">->GetID</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1087 1074
|
|
|
1088 1075 <span class="s">(</span><span class="i">$BondAtomID1</span><span class="cm">,</span> <span class="i">$BondAtomID2</span><span class="s">)</span> = <span class="s">(</span><span class="i">$AtomID</span> < <span class="i">$BondedAtomID</span><span class="s">)</span> ? <span class="s">(</span><span class="i">$AtomID</span><span class="cm">,</span> <span class="i">$BondedAtomID</span><span class="s">)</span> <span class="co">:</span> <span class="s">(</span><span class="i">$BondedAtomID</span><span class="cm">,</span> <span class="i">$AtomID</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1089 1076 <span class="k">push</span> <span class="i">@AtomPathSymbols</span><span class="cm">,</span> <span class="i">$This</span>->{<span class="w">BondSymbols</span>}{<span class="i">$BondAtomID1</span>}{<span class="i">$BondAtomID2</span>}<span class="sc">;</span>
|
|
|
1090 1077
|
|
|
1091 1078 <span class="c"># Process atom type of next atom in path...</span>
|
|
|
1092 1079 <span class="k">push</span> <span class="i">@AtomPathSymbols</span><span class="cm">,</span> <span class="i">$This</span>->{<span class="w">AssignedAtomTypes</span>}{<span class="i">$BondedAtomID</span>}<span class="sc">;</span>
|
|
|
1093 1080 <span class="s">}</span>
|
|
|
1094 1081 <span class="k">return</span> \<span class="i">@AtomPathSymbols</span><span class="sc">;</span>
|
|
|
1095 1082 <span class="s">}</span>
|
|
|
1096 1083
|
|
|
1097 1084 <span class="c"># Set final fingerprits...</span>
|
|
|
1098 1085 <span class="c">#</span>
|
|
|
1099 <a name="_SetFinalFingerprints-"></a>1086 <span class="k">sub </span><span class="m">_SetFinalFingerprints</span> <span class="s">{</span>
|
|
|
1100 1087 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1101 1088
|
|
|
1102 1089 <span class="c"># Mark successful generation of fingerprints...</span>
|
|
|
1103 1090 <span class="i">$This</span>->{<span class="w">FingerprintsGenerated</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
|
1104 1091
|
|
|
1105 1092 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">Type</span>} =~ <span class="q">/^PathLengthBits$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1106 1093 <span class="i">$This</span><span class="i">->_SetFinalFingerprintsBitVector</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1107 1094 <span class="s">}</span>
|
|
|
1108 1095 <span class="k">elsif</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">Type</span>} =~ <span class="q">/^PathLengthCount$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1109 1096 <span class="i">$This</span><span class="i">->_SetFinalFingerprintsVector</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1110 1097 <span class="s">}</span>
|
|
|
1111 1098
|
|
|
1112 1099 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
1113 1100 <span class="s">}</span>
|
|
|
1114 1101
|
|
|
1115 1102 <span class="c"># Set final fingerprits bit vector...</span>
|
|
|
1116 1103 <span class="c">#</span>
|
|
|
1117 <a name="_SetFinalFingerprintsBitVector-"></a>1104 <span class="k">sub </span><span class="m">_SetFinalFingerprintsBitVector</span> <span class="s">{</span>
|
|
|
1118 1105 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1119 1106 <span class="k">my</span><span class="s">(</span><span class="i">$PathLength</span><span class="cm">,</span> <span class="i">$Size</span><span class="cm">,</span> <span class="i">$AtomPathString</span><span class="cm">,</span> <span class="i">$AtomPathHashCode</span><span class="cm">,</span> <span class="i">$AtomPathBitPos</span><span class="cm">,</span> <span class="i">$FingerprintsBitVector</span><span class="cm">,</span> <span class="i">$SkipBitPosCheck</span><span class="cm">,</span> <span class="i">$NumOfBitsToSetPerPath</span><span class="cm">,</span> <span class="i">$SetBitNum</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1120 1107
|
|
|
1121 1108 <span class="i">$FingerprintsBitVector</span> = <span class="i">$This</span>->{<span class="w">FingerprintsBitVector</span>}<span class="sc">;</span>
|
|
|
1122 1109
|
|
|
1123 1110 <span class="i">$Size</span> = <span class="i">$This</span>->{<span class="w">Size</span>}<span class="sc">;</span>
|
|
|
1124 1111
|
|
|
1125 1112 <span class="i">$SkipBitPosCheck</span> = <span class="n">1</span><span class="sc">;</span>
|
|
|
1126 1113 <span class="i">$NumOfBitsToSetPerPath</span> = <span class="i">$This</span>->{<span class="w">NumOfBitsToSetPerPath</span>}<span class="sc">;</span>
|
|
|
1127 1114
|
|
|
1128 1115 <span class="k">for</span> <span class="i">$PathLength</span> <span class="s">(</span><span class="k">keys</span> <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
1129 1116 <span class="k">for</span> <span class="i">$AtomPathString</span> <span class="s">(</span><span class="k">keys</span> <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
1130 1117 <span class="i">$AtomPathHashCode</span> = <span class="i">TextUtil::HashCode</span><span class="s">(</span><span class="i">$AtomPathString</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1131 1118
|
|
|
1132 1119 <span class="c"># Set random number seed...</span>
|
|
|
1133 1120 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">UsePerlCoreRandom</span>}<span class="s">)</span> <span class="s">{</span>
|
|
|
1134 1121 <span class="i">CORE::</span><span class="k">srand</span><span class="s">(</span><span class="i">$AtomPathHashCode</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1135 1122 <span class="s">}</span>
|
|
|
1136 1123 <span class="k">else</span> <span class="s">{</span>
|
|
|
1137 1124 <span class="i">MathUtil::srandom</span><span class="s">(</span><span class="i">$AtomPathHashCode</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1138 1125 <span class="s">}</span>
|
|
|
1139 1126
|
|
|
1140 1127 <span class="k">for</span> <span class="i">$SetBitNum</span> <span class="s">(</span><span class="n">1</span> .. <span class="i">$NumOfBitsToSetPerPath</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1141 1128 <span class="i">$AtomPathBitPos</span> = <span class="i">$This</span>->{<span class="w">UsePerlCoreRandom</span>} ? <span class="k">int</span><span class="s">(</span><span class="i">CORE::</span><span class="k">rand</span><span class="s">(</span><span class="i">$Size</span><span class="s">)</span><span class="s">)</span> <span class="co">:</span> <span class="k">int</span><span class="s">(</span><span class="i">MathUtil::random</span><span class="s">(</span><span class="i">$Size</span><span class="s">)</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1142 1129 <span class="i">$FingerprintsBitVector</span><span class="i">->SetBit</span><span class="s">(</span><span class="i">$AtomPathBitPos</span><span class="cm">,</span> <span class="i">$SkipBitPosCheck</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1143 1130 <span class="s">}</span>
|
|
|
1144 1131 <span class="s">}</span>
|
|
|
1145 1132 <span class="s">}</span>
|
|
|
1146 1133 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
1147 1134 <span class="s">}</span>
|
|
|
1148 1135
|
|
|
1149 1136 <span class="c"># Set final fingerprits vector...</span>
|
|
|
1150 1137 <span class="c">#</span>
|
|
|
1151 <a name="_SetFinalFingerprintsVector-"></a>1138 <span class="k">sub </span><span class="m">_SetFinalFingerprintsVector</span> <span class="s">{</span>
|
|
|
1152 1139 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1153 1140 <span class="k">my</span><span class="s">(</span><span class="i">$PathLength</span><span class="cm">,</span> <span class="i">$AtomPathString</span><span class="cm">,</span> <span class="i">$FingerprintsVector</span><span class="cm">,</span> <span class="i">$AtomPathCount</span><span class="cm">,</span> <span class="i">@Values</span><span class="cm">,</span> <span class="i">@ValueIDs</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1154 1141
|
|
|
1155 1142 <span class="i">@Values</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1156 1143 <span class="i">@ValueIDs</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1157 1144
|
|
|
1158 1145 <span class="k">for</span> <span class="i">$PathLength</span> <span class="s">(</span><span class="k">sort</span> <span class="s">{</span> <span class="i">$a</span> <=> <span class="i">$b</span> <span class="s">}</span> <span class="k">keys</span> <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
1159 1146 <span class="k">for</span> <span class="i">$AtomPathString</span> <span class="s">(</span><span class="k">sort</span> <span class="k">keys</span> <span class="i">%</span>{<span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
|
1160 1147 <span class="i">$AtomPathCount</span> = <span class="i">$This</span>->{<span class="w">AtomPathsStrings</span>}{<span class="i">$PathLength</span>}{<span class="i">$AtomPathString</span>}<span class="sc">;</span>
|
|
|
1161 1148
|
|
|
1162 1149 <span class="k">push</span> <span class="i">@Values</span><span class="cm">,</span> <span class="i">$AtomPathCount</span><span class="sc">;</span>
|
|
|
1163 1150 <span class="k">push</span> <span class="i">@ValueIDs</span><span class="cm">,</span> <span class="i">$AtomPathString</span><span class="sc">;</span>
|
|
|
1164 1151 <span class="s">}</span>
|
|
|
1165 1152 <span class="s">}</span>
|
|
|
1166 1153
|
|
|
1167 1154 <span class="c"># Add PathLengthIDs and values to fingerprint vector...</span>
|
|
|
1168 1155 <span class="i">$This</span>->{<span class="w">FingerprintsVector</span>}<span class="i">->AddValueIDs</span><span class="s">(</span>\<span class="i">@ValueIDs</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1169 1156 <span class="i">$This</span>->{<span class="w">FingerprintsVector</span>}<span class="i">->AddValues</span><span class="s">(</span>\<span class="i">@Values</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1170 1157
|
|
|
1171 1158 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
1172 1159 <span class="s">}</span>
|
|
|
1173 1160
|
|
|
1174 1161 <span class="c"># Cache appropriate molecule data...</span>
|
|
|
1175 1162 <span class="c">#</span>
|
|
|
1176 <a name="_SetupMoleculeDataCache-"></a>1163 <span class="k">sub </span><span class="m">_SetupMoleculeDataCache</span> <span class="s">{</span>
|
|
|
1177 1164 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1178 1165
|
|
|
1179 1166 <span class="c"># Get all atoms...</span>
|
|
|
1180 1167 <span class="i">@</span>{<span class="i">$This</span>->{<span class="w">Atoms</span>}} = <span class="i">$This</span><span class="i">->GetMolecule</span><span class="s">(</span><span class="s">)</span><span class="i">->GetAtoms</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1181 1168
|
|
|
1182 1169 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
1183 1170 <span class="s">}</span>
|
|
|
1184 1171
|
|
|
1185 1172 <span class="c"># Clear cached molecule data...</span>
|
|
|
1186 1173 <span class="c">#</span>
|
|
|
1187 <a name="_ClearMoleculeDataCache-"></a>1174 <span class="k">sub </span><span class="m">_ClearMoleculeDataCache</span> <span class="s">{</span>
|
|
|
1188 1175 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1189 1176
|
|
|
1190 1177 <span class="c"># Clear atoms...</span>
|
|
|
1191 1178 <span class="i">@</span>{<span class="i">$This</span>->{<span class="w">Atoms</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1192 1179
|
|
|
1193 1180 <span class="c"># Clear path atoms..</span>
|
|
|
1194 1181 <span class="i">$This</span>->{<span class="w">AtomPathsRef</span>} = <span class="q">''</span><span class="sc">;</span>
|
|
|
1195 1182
|
|
|
1196 1183 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
1197 1184 <span class="s">}</span>
|
|
|
1198 1185
|
|
|
1199 1186 <span class="c"># Set atomic invariants to use atom identifiers...</span>
|
|
|
1200 1187 <span class="c">#</span>
|
|
|
1201 <a name="SetAtomicInvariantsToUse-"></a>1188 <span class="k">sub </span><span class="m">SetAtomicInvariantsToUse</span> <span class="s">{</span>
|
|
|
1202 1189 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">@Values</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1203 1190 <span class="k">my</span><span class="s">(</span><span class="i">$FirstValue</span><span class="cm">,</span> <span class="i">$TypeOfFirstValue</span><span class="cm">,</span> <span class="i">$AtomicInvariant</span><span class="cm">,</span> <span class="i">$SpecifiedAtomicInvariant</span><span class="cm">,</span> <span class="i">$AtomicInvariantValue</span><span class="cm">,</span> <span class="i">@SpecifiedAtomicInvariants</span><span class="cm">,</span> <span class="i">@AtomicInvariantsToUse</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1204 1191
|
|
|
1205 1192 <span class="k">if</span> <span class="s">(</span>!<span class="i">@Values</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1206 1193 <span class="w">carp</span> <span class="q">"Warning: ${ClassName}->SetAtomicInvariantsToUse: No values specified..."</span><span class="sc">;</span>
|
|
|
1207 1194 <span class="k">return</span><span class="sc">;</span>
|
|
|
1208 1195 <span class="s">}</span>
|
|
|
1209 1196
|
|
|
1210 1197 <span class="i">$FirstValue</span> = <span class="i">$Values</span>[<span class="n">0</span>]<span class="sc">;</span>
|
|
|
1211 1198 <span class="i">$TypeOfFirstValue</span> = <span class="k">ref</span> <span class="i">$FirstValue</span><span class="sc">;</span>
|
|
|
1212 1199
|
|
|
1213 1200 <span class="i">@SpecifiedAtomicInvariants</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1214 1201 <span class="i">@AtomicInvariantsToUse</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1215 1202
|
|
|
1216 1203 <span class="k">if</span> <span class="s">(</span><span class="i">$TypeOfFirstValue</span> =~ <span class="q">/^ARRAY/</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1217 1204 <span class="k">push</span> <span class="i">@SpecifiedAtomicInvariants</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$FirstValue</span>}<span class="sc">;</span>
|
|
|
1218 1205 <span class="s">}</span>
|
|
|
1219 1206 <span class="k">else</span> <span class="s">{</span>
|
|
|
1220 1207 <span class="k">push</span> <span class="i">@SpecifiedAtomicInvariants</span><span class="cm">,</span> <span class="i">@Values</span><span class="sc">;</span>
|
|
|
1221 1208 <span class="s">}</span>
|
|
|
1222 1209
|
|
|
1223 1210 <span class="c"># Make sure specified AtomicInvariants are valid...</span>
|
|
|
1224 1211 <span class="k">for</span> <span class="i">$SpecifiedAtomicInvariant</span> <span class="s">(</span><span class="i">@SpecifiedAtomicInvariants</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1225 1212 <span class="k">if</span> <span class="s">(</span>!<span class="i">AtomTypes::AtomicInvariantsAtomTypes::IsAtomicInvariantAvailable</span><span class="s">(</span><span class="i">$SpecifiedAtomicInvariant</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1226 1213 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->SetAtomicInvariantsToUse: Specified atomic invariant, $SpecifiedAtomicInvariant, is not supported...\n "</span><span class="sc">;</span>
|
|
|
1227 1214 <span class="s">}</span>
|
|
|
1228 1215 <span class="i">$AtomicInvariant</span> = <span class="i">$SpecifiedAtomicInvariant</span><span class="sc">;</span>
|
|
|
1229 1216 <span class="k">push</span> <span class="i">@AtomicInvariantsToUse</span><span class="cm">,</span> <span class="i">$AtomicInvariant</span><span class="sc">;</span>
|
|
|
1230 1217 <span class="s">}</span>
|
|
|
1231 1218
|
|
|
1232 1219 <span class="c"># Set atomic invariants to use...</span>
|
|
|
1233 1220 <span class="i">@</span>{<span class="i">$This</span>->{<span class="w">AtomicInvariantsToUse</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1234 1221 <span class="k">push</span> <span class="i">@</span>{<span class="i">$This</span>->{<span class="w">AtomicInvariantsToUse</span>}}<span class="cm">,</span> <span class="i">@AtomicInvariantsToUse</span><span class="sc">;</span>
|
|
|
1235 1222
|
|
|
1236 1223 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
1237 1224 <span class="s">}</span>
|
|
|
1238 1225
|
|
|
1239 1226 <span class="c"># Set functional classes to use for atom identifiers...</span>
|
|
|
1240 1227 <span class="c">#</span>
|
|
|
1241 <a name="SetFunctionalClassesToUse-"></a>1228 <span class="k">sub </span><span class="m">SetFunctionalClassesToUse</span> <span class="s">{</span>
|
|
|
1242 1229 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="cm">,</span> <span class="i">@Values</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1243 1230 <span class="k">my</span><span class="s">(</span><span class="i">$FirstValue</span><span class="cm">,</span> <span class="i">$TypeOfFirstValue</span><span class="cm">,</span> <span class="i">$FunctionalClass</span><span class="cm">,</span> <span class="i">$SpecifiedFunctionalClass</span><span class="cm">,</span> <span class="i">@SpecifiedFunctionalClasses</span><span class="cm">,</span> <span class="i">@FunctionalClassesToUse</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1244 1231
|
|
|
1245 1232 <span class="k">if</span> <span class="s">(</span>!<span class="i">@Values</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1246 1233 <span class="w">carp</span> <span class="q">"Warning: ${ClassName}->SetFunctionalClassesToUse: No values specified..."</span><span class="sc">;</span>
|
|
|
1247 1234 <span class="k">return</span><span class="sc">;</span>
|
|
|
1248 1235 <span class="s">}</span>
|
|
|
1249 1236
|
|
|
1250 1237 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} !~ <span class="q">/^FunctionalClassAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1251 1238 <span class="w">carp</span> <span class="q">"Warning: ${ClassName}->SetFunctionalClassesToUse: FunctionalClassesToUse can't be set for InitialAtomIdentifierType of $This->{AtomIdentifierType}..."</span><span class="sc">;</span>
|
|
|
1252 1239 <span class="k">return</span><span class="sc">;</span>
|
|
|
1253 1240 <span class="s">}</span>
|
|
|
1254 1241
|
|
|
1255 1242 <span class="i">$FirstValue</span> = <span class="i">$Values</span>[<span class="n">0</span>]<span class="sc">;</span>
|
|
|
1256 1243 <span class="i">$TypeOfFirstValue</span> = <span class="k">ref</span> <span class="i">$FirstValue</span><span class="sc">;</span>
|
|
|
1257 1244
|
|
|
1258 1245 <span class="i">@SpecifiedFunctionalClasses</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1259 1246 <span class="i">@FunctionalClassesToUse</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1260 1247
|
|
|
1261 1248 <span class="k">if</span> <span class="s">(</span><span class="i">$TypeOfFirstValue</span> =~ <span class="q">/^ARRAY/</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1262 1249 <span class="k">push</span> <span class="i">@SpecifiedFunctionalClasses</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$FirstValue</span>}<span class="sc">;</span>
|
|
|
1263 1250 <span class="s">}</span>
|
|
|
1264 1251 <span class="k">else</span> <span class="s">{</span>
|
|
|
1265 1252 <span class="k">push</span> <span class="i">@SpecifiedFunctionalClasses</span><span class="cm">,</span> <span class="i">@Values</span><span class="sc">;</span>
|
|
|
1266 1253 <span class="s">}</span>
|
|
|
1267 1254
|
|
|
1268 1255 <span class="c"># Make sure specified FunctionalClasses are valid...</span>
|
|
|
1269 1256 <span class="k">for</span> <span class="i">$SpecifiedFunctionalClass</span> <span class="s">(</span><span class="i">@SpecifiedFunctionalClasses</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1270 1257 <span class="k">if</span> <span class="s">(</span>!<span class="i">AtomTypes::FunctionalClassAtomTypes::IsFunctionalClassAvailable</span><span class="s">(</span><span class="i">$SpecifiedFunctionalClass</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1271 1258 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->SetFunctionalClassesToUse: Specified functional class, $SpecifiedFunctionalClass, is not supported...\n "</span><span class="sc">;</span>
|
|
|
1272 1259 <span class="s">}</span>
|
|
|
1273 1260 <span class="k">push</span> <span class="i">@FunctionalClassesToUse</span><span class="cm">,</span> <span class="i">$SpecifiedFunctionalClass</span><span class="sc">;</span>
|
|
|
1274 1261 <span class="s">}</span>
|
|
|
1275 1262
|
|
|
1276 1263 <span class="c"># Set functional classes to use...</span>
|
|
|
1277 1264 <span class="i">@</span>{<span class="i">$This</span>->{<span class="w">FunctionalClassesToUse</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1278 1265 <span class="k">push</span> <span class="i">@</span>{<span class="i">$This</span>->{<span class="w">FunctionalClassesToUse</span>}}<span class="cm">,</span> <span class="i">@FunctionalClassesToUse</span><span class="sc">;</span>
|
|
|
1279 1266
|
|
|
1280 1267 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
1281 1268 <span class="s">}</span>
|
|
|
1282 1269
|
|
|
1283 1270 <span class="c"># Initialize atom indentifier type information...</span>
|
|
|
1284 1271 <span class="c">#</span>
|
|
|
1285 1272 <span class="c"># Current supported values:</span>
|
|
|
1286 1273 <span class="c">#</span>
|
|
|
1287 1274 <span class="c"># AtomicInvariantsAtomTypes, DREIDINGAtomTypes, EStateAtomTypes, FunctionalClassAtomTypes,</span>
|
|
|
1288 1275 <span class="c"># MMFF94AtomTypes, SLogPAtomTypes, SYBYLAtomTypes, TPSAAtomTypes, UFFAtomTypes</span>
|
|
|
1289 1276 <span class="c">#</span>
|
|
|
1290 <a name="_InitializeAtomIdentifierTypeInformation-"></a>1277 <span class="k">sub </span><span class="m">_InitializeAtomIdentifierTypeInformation</span> <span class="s">{</span>
|
|
|
1291 1278 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1292 1279
|
|
|
1293 1280 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^AtomicInvariantsAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1294 1281 <span class="i">$This</span><span class="i">->_InitializeAtomicInvariantsAtomTypesInformation</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1295 1282 <span class="s">}</span>
|
|
|
1296 1283 <span class="k">elsif</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^FunctionalClassAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1297 1284 <span class="i">$This</span><span class="i">->_InitializeFunctionalClassAtomTypesInformation</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1298 1285 <span class="s">}</span>
|
|
|
1299 1286 <span class="k">elsif</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^(DREIDINGAtomTypes|EStateAtomTypes|MMFF94AtomTypes|SLogPAtomTypes|SYBYLAtomTypes|TPSAAtomTypes|UFFAtomTypes)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1300 1287 <span class="c"># Nothing to do for now...</span>
|
|
|
1301 1288 <span class="s">}</span>
|
|
|
1302 1289 <span class="k">else</span> <span class="s">{</span>
|
|
|
1303 1290 <span class="w">croak</span> <span class="q">"Error: ${ClassName}->_InitializeAtomIdentifierTypeInformation: Unknown atom indentifier type $This->{AtomIdentifierType}..."</span><span class="sc">;</span>
|
|
|
1304 1291 <span class="s">}</span>
|
|
|
1305 1292
|
|
|
1306 1293 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
1307 1294 <span class="s">}</span>
|
|
|
1308 1295
|
|
|
1309 1296 <span class="c"># Initialize atomic invariants atom types to use for generating atom identifiers...</span>
|
|
|
1310 1297 <span class="c">#</span>
|
|
|
1311 1298 <span class="c"># Let:</span>
|
|
|
1312 1299 <span class="c"># AS = Atom symbol corresponding to element symbol</span>
|
|
|
1313 1300 <span class="c">#</span>
|
|
|
1314 1301 <span class="c"># X<n> = Number of non-hydrogen atom neighbors or heavy atoms attached to atom</span>
|
|
|
1315 1302 <span class="c"># BO<n> = Sum of bond orders to non-hydrogen atom neighbors or heavy atoms attached to atom</span>
|
|
|
1316 1303 <span class="c"># LBO<n> = Largest bond order of non-hydrogen atom neighbors or heavy atoms attached to atom</span>
|
|
|
1317 1304 <span class="c"># SB<n> = Number of single bonds to non-hydrogen atom neighbors or heavy atoms attached to atom</span>
|
|
|
1318 1305 <span class="c"># DB<n> = Number of double bonds to non-hydrogen atom neighbors or heavy atoms attached to atom</span>
|
|
|
1319 1306 <span class="c"># TB<n> = Number of triple bonds to non-hydrogen atom neighbors or heavy atoms attached to atom</span>
|
|
|
1320 1307 <span class="c"># H<n> = Number of implicit and explicit hydrogens for atom</span>
|
|
|
1321 1308 <span class="c"># Ar = Aromatic annotation indicating whether atom is aromatic</span>
|
|
|
1322 1309 <span class="c"># RA = Ring atom annotation indicating whether atom is a ring</span>
|
|
|
1323 1310 <span class="c"># FC<+n/-n> = Formal charge assigned to atom</span>
|
|
|
1324 1311 <span class="c"># MN<n> = Mass number indicating isotope other than most abundant isotope</span>
|
|
|
1325 1312 <span class="c"># SM<n> = Spin multiplicity of atom. Possible values: 1 (singlet), 2 (doublet) or 3 (triplet)</span>
|
|
|
1326 1313 <span class="c">#</span>
|
|
|
1327 1314 <span class="c"># Then:</span>
|
|
|
1328 1315 <span class="c">#</span>
|
|
|
1329 1316 <span class="c"># Atom type generated by AtomTypes::AtomicInvariantsAtomTypes class corresponds to:</span>
|
|
|
1330 1317 <span class="c">#</span>
|
|
|
1331 1318 <span class="c"># AS.X<n>.BO<n>.LBO<n>.<SB><n>.<DB><n>.<TB><n>.H<n>.Ar.RA.FC<+n/-n>.MN<n>.SM<n></span>
|
|
|
1332 1319 <span class="c">#</span>
|
|
|
1333 1320 <span class="c"># Except for AS which is a required atomic invariant in atom types, all other atomic invariants are</span>
|
|
|
1334 1321 <span class="c"># optional. Default atomic invariants used for AtomID are: AS, X<n>, BO<n>, H<n>, FC<+n/-n>.</span>
|
|
|
1335 1322 <span class="c"># AtomID specification doesn't include atomic invariants with zero or undefined values.</span>
|
|
|
1336 1323 <span class="c">#</span>
|
|
|
1337 <a name="_InitializeAtomicInvariantsAtomTypesInformation-"></a>1324 <span class="k">sub </span><span class="m">_InitializeAtomicInvariantsAtomTypesInformation</span> <span class="s">{</span>
|
|
|
1338 1325 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1339 1326
|
|
|
1340 1327 <span class="c"># Default atomic invariants to use for generating atom neighborhood atom IDs: AS, X, BO, H, FC</span>
|
|
|
1341 1328 <span class="c">#</span>
|
|
|
1342 1329 <span class="i">@</span>{<span class="i">$This</span>->{<span class="w">AtomicInvariantsToUse</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1343 1330 <span class="i">@</span>{<span class="i">$This</span>->{<span class="w">AtomicInvariantsToUse</span>}} = <span class="s">(</span><span class="q">'AS'</span><span class="cm">,</span> <span class="q">'X'</span><span class="cm">,</span> <span class="q">'BO'</span><span class="cm">,</span> <span class="q">'H'</span><span class="cm">,</span> <span class="q">'FC'</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1344 1331
|
|
|
1345 1332 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
1346 1333 <span class="s">}</span>
|
|
|
1347 1334
|
|
|
1348 1335 <span class="c"># Initialize functional class atom types, generated by AtomTypes::FunctionalClassAtomTypes</span>
|
|
|
1349 1336 <span class="c"># class, to use for generating atom identifiers...</span>
|
|
|
1350 1337 <span class="c">#</span>
|
|
|
1351 1338 <span class="c"># Let:</span>
|
|
|
1352 1339 <span class="c"># HBD: HydrogenBondDonor</span>
|
|
|
1353 1340 <span class="c"># HBA: HydrogenBondAcceptor</span>
|
|
|
1354 1341 <span class="c"># PI : PositivelyIonizable</span>
|
|
|
1355 1342 <span class="c"># NI : NegativelyIonizable</span>
|
|
|
1356 1343 <span class="c"># Ar : Aromatic</span>
|
|
|
1357 1344 <span class="c"># Hal : Halogen</span>
|
|
|
1358 1345 <span class="c"># H : Hydrophobic</span>
|
|
|
1359 1346 <span class="c"># RA : RingAtom</span>
|
|
|
1360 1347 <span class="c"># CA : ChainAtom</span>
|
|
|
1361 1348 <span class="c">#</span>
|
|
|
1362 1349 <span class="c"># Then:</span>
|
|
|
1363 1350 <span class="c">#</span>
|
|
|
1364 1351 <span class="c"># Functiononal class atom type specification for an atom corresponds to:</span>
|
|
|
1365 1352 <span class="c">#</span>
|
|
|
1366 1353 <span class="c"># Ar.CA.H.HBA.HBD.Hal.NI.PI.RA</span>
|
|
|
1367 1354 <span class="c">#</span>
|
|
|
1368 1355 <span class="c"># Default functional classes used are: HBD, HBA, PI, NI, Ar, Hal</span>
|
|
|
1369 1356 <span class="c">#</span>
|
|
|
1370 1357 <span class="c"># FunctionalAtomTypes are assigned using the following definitions [ Ref 60-61, Ref 65-66 ]:</span>
|
|
|
1371 1358 <span class="c">#</span>
|
|
|
1372 1359 <span class="c"># HydrogenBondDonor: NH, NH2, OH</span>
|
|
|
1373 1360 <span class="c"># HydrogenBondAcceptor: N[!H], O</span>
|
|
|
1374 1361 <span class="c"># PositivelyIonizable: +, NH2</span>
|
|
|
1375 1362 <span class="c"># NegativelyIonizable: -, C(=O)OH, S(=O)OH, P(=O)OH</span>
|
|
|
1376 1363 <span class="c">#</span>
|
|
|
1377 <a name="_InitializeFunctionalClassAtomTypesInformation-"></a>1364 <span class="k">sub </span><span class="m">_InitializeFunctionalClassAtomTypesInformation</span> <span class="s">{</span>
|
|
|
1378 1365 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1379 1366
|
|
|
1380 1367 <span class="c"># Default functional class atom typess to use for generating atom identifiers</span>
|
|
|
1381 1368 <span class="c"># are: HBD, HBA, PI, NI, Ar, Hal</span>
|
|
|
1382 1369 <span class="c">#</span>
|
|
|
1383 1370 <span class="i">@</span>{<span class="i">$This</span>->{<span class="w">FunctionalClassesToUse</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1384 1371 <span class="i">@</span>{<span class="i">$This</span>->{<span class="w">FunctionalClassesToUse</span>}} = <span class="s">(</span><span class="q">'HBD'</span><span class="cm">,</span> <span class="q">'HBA'</span><span class="cm">,</span> <span class="q">'PI'</span><span class="cm">,</span> <span class="q">'NI'</span><span class="cm">,</span> <span class="q">'Ar'</span><span class="cm">,</span> <span class="q">'Hal'</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1385 1372
|
|
|
1386 1373 <span class="k">return</span> <span class="i">$This</span><span class="sc">;</span>
|
|
|
1387 1374 <span class="s">}</span>
|
|
|
1388 1375
|
|
|
1389 1376 <span class="c"># Return a string containg data for PathLengthFingerprints object...</span>
|
|
|
1390 1377 <span class="c">#</span>
|
|
|
1391 <a name="StringifyPathLengthFingerprints-"></a>1378 <span class="k">sub </span><span class="m">StringifyPathLengthFingerprints</span> <span class="s">{</span>
|
|
|
1392 1379 <span class="k">my</span><span class="s">(</span><span class="i">$This</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
|
1393 1380 <span class="k">my</span><span class="s">(</span><span class="i">$PathLengthsFingerprintsString</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1394 1381
|
|
|
1395 1382 <span class="c"># Type of fingerprint...</span>
|
|
|
1396 1383 <span class="i">$PathLengthsFingerprintsString</span> = <span class="q">"Fingerprint type: $This->{Type}; AtomIdentifierType: $This->{AtomIdentifierType}"</span><span class="sc">;</span>
|
|
|
1397 1384
|
|
|
1398 1385 <span class="c"># Path length...</span>
|
|
|
1399 1386 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; MinPathLength: $This->{MinLength}; MaxPathLength: $This->{MaxLength}"</span><span class="sc">;</span>
|
|
|
1400 1387
|
|
|
1401 1388 <span class="c"># Fingerprint generation control...</span>
|
|
|
1402 1389 <span class="k">my</span><span class="s">(</span><span class="i">$AllowSharedBonds</span><span class="cm">,</span> <span class="i">$AllowRings</span><span class="cm">,</span> <span class="i">$UseBondSymbols</span><span class="cm">,</span> <span class="i">$UseUniquePaths</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1403 1390
|
|
|
1404 1391 <span class="i">$AllowSharedBonds</span> = <span class="i">$This</span>->{<span class="w">AllowSharedBonds</span>} ? <span class="q">"Yes"</span> <span class="co">:</span> <span class="q">"No"</span><span class="sc">;</span>
|
|
|
1405 1392 <span class="i">$AllowRings</span> = <span class="i">$This</span>->{<span class="w">AllowRings</span>} ? <span class="q">"Yes"</span> <span class="co">:</span> <span class="q">"No"</span><span class="sc">;</span>
|
|
|
1406 1393 <span class="i">$UseBondSymbols</span> = <span class="i">$This</span>->{<span class="w">UseBondSymbols</span>} ? <span class="q">"Yes"</span> <span class="co">:</span> <span class="q">"No"</span><span class="sc">;</span>
|
|
|
1407 1394 <span class="i">$UseUniquePaths</span> = <span class="i">$This</span>->{<span class="w">UseBondSymbols</span>} ? <span class="q">"Yes"</span> <span class="co">:</span> <span class="q">"No"</span><span class="sc">;</span>
|
|
|
1408 1395
|
|
|
1409 1396 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; UseUniquePaths: $UseUniquePaths; AllowSharedBonds: $AllowSharedBonds; AllowRings: $AllowRings; UseBondSymbols: $UseBondSymbols"</span><span class="sc">;</span>
|
|
|
1410 1397
|
|
|
1411 1398 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^AtomicInvariantsAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1412 1399 <span class="k">my</span><span class="s">(</span><span class="i">$AtomicInvariant</span><span class="cm">,</span> <span class="i">@AtomicInvariants</span><span class="cm">,</span> <span class="i">@AtomicInvariantsOrder</span><span class="cm">,</span> <span class="i">%AvailableAtomicInvariants</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1413 1400
|
|
|
1414 1401 <span class="i">@AtomicInvariantsOrder</span> = <span class="i">AtomTypes::AtomicInvariantsAtomTypes::GetAtomicInvariantsOrder</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1415 1402 <span class="i">%AvailableAtomicInvariants</span> = <span class="i">AtomTypes::AtomicInvariantsAtomTypes::GetAvailableAtomicInvariants</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1416 1403
|
|
|
1417 1404 <span class="k">for</span> <span class="i">$AtomicInvariant</span> <span class="s">(</span><span class="i">@AtomicInvariantsOrder</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1418 1405 <span class="k">push</span> <span class="i">@AtomicInvariants</span><span class="cm">,</span> <span class="q">"$AtomicInvariant: $AvailableAtomicInvariants{$AtomicInvariant}"</span><span class="sc">;</span>
|
|
|
1419 1406 <span class="s">}</span>
|
|
|
1420 1407
|
|
|
1421 1408 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; AtomicInvariantsToUse: <"</span> . <span class="i">TextUtil::JoinWords</span><span class="s">(</span>\<span class="i">@</span>{<span class="i">$This</span>->{<span class="w">AtomicInvariantsToUse</span>}}<span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span> . <span class="q">">"</span><span class="sc">;</span>
|
|
|
1422 1409 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; AtomicInvariantsOrder: <"</span> . <span class="i">TextUtil::JoinWords</span><span class="s">(</span>\<span class="i">@AtomicInvariantsOrder</span><span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span> . <span class="q">">"</span><span class="sc">;</span>
|
|
|
1423 1410 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; AvailableAtomicInvariants: <"</span> . <span class="i">TextUtil::JoinWords</span><span class="s">(</span>\<span class="i">@AtomicInvariants</span><span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span> . <span class="q">">"</span><span class="sc">;</span>
|
|
|
1424 1411 <span class="s">}</span>
|
|
|
1425 1412 <span class="k">elsif</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">AtomIdentifierType</span>} =~ <span class="q">/^FunctionalClassAtomTypes$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1426 1413 <span class="k">my</span><span class="s">(</span><span class="i">$FunctionalClass</span><span class="cm">,</span> <span class="i">@FunctionalClasses</span><span class="cm">,</span> <span class="i">@FunctionalClassesOrder</span><span class="cm">,</span> <span class="i">%AvailableFunctionalClasses</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1427 1414
|
|
|
1428 1415 <span class="i">@FunctionalClassesOrder</span> = <span class="i">AtomTypes::FunctionalClassAtomTypes::GetFunctionalClassesOrder</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1429 1416 <span class="i">%AvailableFunctionalClasses</span> = <span class="i">AtomTypes::FunctionalClassAtomTypes::GetAvailableFunctionalClasses</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1430 1417
|
|
|
1431 1418 <span class="k">for</span> <span class="i">$FunctionalClass</span> <span class="s">(</span><span class="i">@FunctionalClassesOrder</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1432 1419 <span class="k">push</span> <span class="i">@FunctionalClasses</span><span class="cm">,</span> <span class="q">"$FunctionalClass: $AvailableFunctionalClasses{$FunctionalClass}"</span><span class="sc">;</span>
|
|
|
1433 1420 <span class="s">}</span>
|
|
|
1434 1421
|
|
|
1435 1422 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; FunctionalClassesToUse: <"</span> . <span class="i">TextUtil::JoinWords</span><span class="s">(</span>\<span class="i">@</span>{<span class="i">$This</span>->{<span class="w">FunctionalClassesToUse</span>}}<span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span> . <span class="q">">"</span><span class="sc">;</span>
|
|
|
1436 1423 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; FunctionalClassesOrder: <"</span> . <span class="i">TextUtil::JoinWords</span><span class="s">(</span>\<span class="i">@FunctionalClassesOrder</span><span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span> . <span class="q">">"</span><span class="sc">;</span>
|
|
|
1437 1424 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; AvailableFunctionalClasses: <"</span> . <span class="i">TextUtil::JoinWords</span><span class="s">(</span>\<span class="i">@FunctionalClasses</span><span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span> . <span class="q">">"</span><span class="sc">;</span>
|
|
|
1438 1425 <span class="s">}</span>
|
|
|
1439 1426
|
|
|
1440 1427 <span class="k">if</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">Type</span>} =~ <span class="q">/^PathLengthBits$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1441 1428 <span class="c"># Size...</span>
|
|
|
1442 1429 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; Size: $This->{Size}; MinSize: $This->{MinSize}; MaxSize: $This->{MaxSize}"</span><span class="sc">;</span>
|
|
|
1443 1430
|
|
|
1444 1431 <span class="c"># NumOfBitsToSetPerPath...</span>
|
|
|
1445 1432 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; NumOfBitsToSetPerPath: $This->{NumOfBitsToSetPerPath}"</span><span class="sc">;</span>
|
|
|
1446 1433
|
|
|
1447 1434 <span class="c"># Fingerprint bit density and num of bits set...</span>
|
|
|
1448 1435 <span class="k">my</span><span class="s">(</span><span class="i">$NumOfSetBits</span><span class="cm">,</span> <span class="i">$BitDensity</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1449 1436 <span class="i">$NumOfSetBits</span> = <span class="i">$This</span>->{<span class="w">FingerprintsBitVector</span>}<span class="i">->GetNumOfSetBits</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1450 1437 <span class="i">$BitDensity</span> = <span class="i">$This</span>->{<span class="w">FingerprintsBitVector</span>}<span class="i">->GetFingerprintsBitDensity</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
|
1451 1438 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; NumOfOnBits: $NumOfSetBits; BitDensity: $BitDensity"</span><span class="sc">;</span>
|
|
|
1452 1439
|
|
|
1453 1440 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; FingerprintsBitVector: < $This->{FingerprintsBitVector} >"</span><span class="sc">;</span>
|
|
|
1454 1441 <span class="s">}</span>
|
|
|
1455 1442 <span class="k">elsif</span> <span class="s">(</span><span class="i">$This</span>->{<span class="w">Type</span>} =~ <span class="q">/^PathLengthCount$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
|
1456 1443 <span class="i">$PathLengthsFingerprintsString</span> .= <span class="q">"; FingerprintsVector: < $This->{FingerprintsVector} >"</span><span class="sc">;</span>
|
|
|
1457 1444 <span class="s">}</span>
|
|
|
1458 1445
|
|
|
1459 1446 <span class="k">return</span> <span class="i">$PathLengthsFingerprintsString</span><span class="sc">;</span>
|
|
|
1460 1447 <span class="s">}</span>
|
|
|
1461 1448
|
|
|
1462 <a name="EOF-"></a></pre>
|
|
|
1463 <p> </p>
|
|
|
1464 <br />
|
|
|
1465 <center>
|
|
|
1466 <img src="../../../images/h2o2.png">
|
|
|
1467 </center>
|
|
|
1468 </body>
|
|
|
1469 </html>
|
