rino 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- data/README +44 -0
- data/Rakefile +123 -0
- data/ext/extconf.rb +26 -0
- data/ext/ruby_inchi_main.so +0 -0
- data/ext/src/aux2atom.h +2786 -0
- data/ext/src/comdef.h +148 -0
- data/ext/src/e_0dstereo.c +3014 -0
- data/ext/src/e_0dstereo.h +31 -0
- data/ext/src/e_comdef.h +57 -0
- data/ext/src/e_ctl_data.h +147 -0
- data/ext/src/e_ichi_io.c +498 -0
- data/ext/src/e_ichi_io.h +40 -0
- data/ext/src/e_ichi_parms.c +37 -0
- data/ext/src/e_ichi_parms.h +41 -0
- data/ext/src/e_ichicomp.h +50 -0
- data/ext/src/e_ichierr.h +40 -0
- data/ext/src/e_ichimain.c +593 -0
- data/ext/src/e_ichisize.h +43 -0
- data/ext/src/e_inchi_atom.c +75 -0
- data/ext/src/e_inchi_atom.h +33 -0
- data/ext/src/e_inpdef.h +41 -0
- data/ext/src/e_mode.h +706 -0
- data/ext/src/e_mol2atom.c +649 -0
- data/ext/src/e_readinch.c +58 -0
- data/ext/src/e_readmol.c +54 -0
- data/ext/src/e_readmol.h +180 -0
- data/ext/src/e_readstru.c +251 -0
- data/ext/src/e_readstru.h +33 -0
- data/ext/src/e_util.c +284 -0
- data/ext/src/e_util.h +61 -0
- data/ext/src/extr_ct.h +251 -0
- data/ext/src/ichi.h +206 -0
- data/ext/src/ichi_bns.c +7999 -0
- data/ext/src/ichi_bns.h +231 -0
- data/ext/src/ichican2.c +5000 -0
- data/ext/src/ichicano.c +2195 -0
- data/ext/src/ichicano.h +49 -0
- data/ext/src/ichicans.c +1625 -0
- data/ext/src/ichicant.h +379 -0
- data/ext/src/ichicomn.h +260 -0
- data/ext/src/ichicomp.h +50 -0
- data/ext/src/ichidrp.h +119 -0
- data/ext/src/ichierr.h +124 -0
- data/ext/src/ichiisot.c +101 -0
- data/ext/src/ichilnct.c +286 -0
- data/ext/src/ichimain.h +132 -0
- data/ext/src/ichimak2.c +1189 -0
- data/ext/src/ichimake.c +3812 -0
- data/ext/src/ichimake.h +205 -0
- data/ext/src/ichimap1.c +851 -0
- data/ext/src/ichimap2.c +2856 -0
- data/ext/src/ichimap4.c +1609 -0
- data/ext/src/ichinorm.c +741 -0
- data/ext/src/ichinorm.h +67 -0
- data/ext/src/ichiparm.c +45 -0
- data/ext/src/ichiparm.h +1441 -0
- data/ext/src/ichiprt1.c +3612 -0
- data/ext/src/ichiprt2.c +1511 -0
- data/ext/src/ichiprt3.c +3011 -0
- data/ext/src/ichiqueu.c +1003 -0
- data/ext/src/ichiring.c +326 -0
- data/ext/src/ichiring.h +49 -0
- data/ext/src/ichisize.h +35 -0
- data/ext/src/ichisort.c +539 -0
- data/ext/src/ichister.c +3538 -0
- data/ext/src/ichister.h +35 -0
- data/ext/src/ichitaut.c +3843 -0
- data/ext/src/ichitaut.h +387 -0
- data/ext/src/ichitime.h +74 -0
- data/ext/src/inchi_api.h +670 -0
- data/ext/src/inchi_dll.c +1480 -0
- data/ext/src/inchi_dll.h +34 -0
- data/ext/src/inchi_dll_main.c +23 -0
- data/ext/src/inchi_dll_main.h +31 -0
- data/ext/src/inpdef.h +328 -0
- data/ext/src/lreadmol.h +1246 -0
- data/ext/src/mode.h +706 -0
- data/ext/src/ruby_inchi_main.c +558 -0
- data/ext/src/runichi.c +4179 -0
- data/ext/src/strutil.c +3861 -0
- data/ext/src/strutil.h +182 -0
- data/ext/src/util.c +1130 -0
- data/ext/src/util.h +85 -0
- data/lib/clean_tempfile.rb +220 -0
- data/lib/rino.rb +111 -0
- data/test/test.rb +386 -0
- metadata +130 -0
data/ext/src/inchi_api.h
ADDED
@@ -0,0 +1,670 @@
|
|
1
|
+
/*
|
2
|
+
* International Union of Pure and Applied Chemistry (IUPAC)
|
3
|
+
* International Chemical Identifier (InChI)
|
4
|
+
* Version 1
|
5
|
+
* Software version 1.00
|
6
|
+
* April 13, 2005
|
7
|
+
* Developed at NIST
|
8
|
+
*/
|
9
|
+
|
10
|
+
#ifndef __INHCH_API_H__
|
11
|
+
#define __INHCH_API_H__
|
12
|
+
|
13
|
+
/* radical definitions */
|
14
|
+
typedef enum tagINCHIRadical {
|
15
|
+
INCHI_RADICAL_NONE = 0,
|
16
|
+
INCHI_RADICAL_SINGLET = 1,
|
17
|
+
INCHI_RADICAL_DOUBLET = 2,
|
18
|
+
INCHI_RADICAL_TRIPLET = 3
|
19
|
+
} inchi_Radical;
|
20
|
+
|
21
|
+
/* bond type definitions */
|
22
|
+
typedef enum tagINCHIBondType {
|
23
|
+
INCHI_BOND_TYPE_NONE = 0,
|
24
|
+
INCHI_BOND_TYPE_SINGLE = 1,
|
25
|
+
INCHI_BOND_TYPE_DOUBLE = 2,
|
26
|
+
INCHI_BOND_TYPE_TRIPLE = 3,
|
27
|
+
INCHI_BOND_TYPE_ALTERN = 4
|
28
|
+
} inchi_BondType;
|
29
|
+
/* 2D stereo definitions */
|
30
|
+
typedef enum tagINCHIBondStereo2D {
|
31
|
+
/* stereocenter-related; positive: the sharp end points to this atom */
|
32
|
+
INCHI_BOND_STEREO_NONE = 0,
|
33
|
+
INCHI_BOND_STEREO_SINGLE_1UP = 1,
|
34
|
+
INCHI_BOND_STEREO_SINGLE_1EITHER = 4,
|
35
|
+
INCHI_BOND_STEREO_SINGLE_1DOWN = 6,
|
36
|
+
/* stereocenter-related; negative: the sharp end points to the opposite atom */
|
37
|
+
INCHI_BOND_STEREO_SINGLE_2UP = -1,
|
38
|
+
INCHI_BOND_STEREO_SINGLE_2EITHER = -4,
|
39
|
+
INCHI_BOND_STEREO_SINGLE_2DOWN = -6,
|
40
|
+
/* stereobond-related */
|
41
|
+
INCHI_BOND_STEREO_DOUBLE_EITHER = 3 /* unknown stereobond geometry */
|
42
|
+
} inchi_BondStereo2D;
|
43
|
+
|
44
|
+
/*************************************************************************
|
45
|
+
* Notes on using INCHI_BOND_STEREO_SINGLE_* from inchi_BondStereo2D *
|
46
|
+
* *
|
47
|
+
* These stereo markings are used by InChI to characterize a stereogenic *
|
48
|
+
* atom if and only if all neighbors of this atom have same z-coordinate *
|
49
|
+
* as this atom (that is, in case of 2D fragment). *
|
50
|
+
* The only exception is INCHI_BOND_STEREO_SINGLE_?EITHER marking which *
|
51
|
+
* always assigns to the atom an "unknown" parity (u). *
|
52
|
+
* *
|
53
|
+
* Note that the -NEWPS InChI option changes 2D stereo interpretation: *
|
54
|
+
* only bonds that have sharp end pointing to the stereogenic atom are *
|
55
|
+
* considered as being out of plane and only sharp ends of *
|
56
|
+
* INCHI_BOND_STEREO_SINGLE_?EITHER bonds are considered to determine *
|
57
|
+
* whether the stereochemistry is unknown. *
|
58
|
+
*************************************************************************/
|
59
|
+
|
60
|
+
/* sizes definitions */
|
61
|
+
#define MAXVAL 20 /* max number of bonds per atom */
|
62
|
+
#define ATOM_EL_LEN 6 /* length of ASCIIZ element symbol field */
|
63
|
+
#define NUM_H_ISOTOPES 3 /* number of hydrogen isotopes: protium, D, T */
|
64
|
+
#define ISOTOPIC_SHIFT_FLAG 10000 /* add to isotopic mass if isotopic_mass = */
|
65
|
+
/* (isotopic mass - average atomic mass) */
|
66
|
+
#define ISOTOPIC_SHIFT_MAX 100 /* max abs(isotopic mass - average atomic mass) */
|
67
|
+
|
68
|
+
#ifndef INCHI_US_CHAR_DEF
|
69
|
+
typedef signed char S_CHAR;
|
70
|
+
typedef unsigned char U_CHAR;
|
71
|
+
#define INCHI_US_CHAR_DEF
|
72
|
+
#endif
|
73
|
+
|
74
|
+
#ifndef INCHI_US_SHORT_DEF
|
75
|
+
typedef signed short S_SHORT;
|
76
|
+
typedef unsigned short U_SHORT;
|
77
|
+
#define INCHI_US_SHORT_DEF
|
78
|
+
#endif
|
79
|
+
|
80
|
+
typedef S_SHORT AT_NUM; /* atom number; starts from 0 */
|
81
|
+
|
82
|
+
/*************************************************
|
83
|
+
*
|
84
|
+
*
|
85
|
+
* A T O M S a n d C O N N E C T I V I T Y
|
86
|
+
*
|
87
|
+
*
|
88
|
+
*************************************************/
|
89
|
+
|
90
|
+
typedef struct tagInchiAtom {
|
91
|
+
/* atom coordinates */
|
92
|
+
double x;
|
93
|
+
double y;
|
94
|
+
double z;
|
95
|
+
/* connectivity */
|
96
|
+
AT_NUM neighbor[MAXVAL]; /* adjacency list: ordering numbers of */
|
97
|
+
/* the adjacent atoms, >= 0 */
|
98
|
+
S_CHAR bond_type[MAXVAL]; /* inchi_BondType */
|
99
|
+
/* 2D stereo */
|
100
|
+
S_CHAR bond_stereo[MAXVAL]; /* inchi_BondStereo2D; negative if the */
|
101
|
+
/* sharp end points to opposite atom */
|
102
|
+
/* other atom properties */
|
103
|
+
char elname[ATOM_EL_LEN]; /* zero-terminated chemical element name:*/
|
104
|
+
/* "H", "Si", etc. */
|
105
|
+
AT_NUM num_bonds; /* number of neighbors, bond types and bond*/
|
106
|
+
/* stereo in the adjacency list */
|
107
|
+
S_CHAR num_iso_H[NUM_H_ISOTOPES+1]; /* implicit hydrogen atoms */
|
108
|
+
/* [0]: number of implicit non-isotopic H
|
109
|
+
(exception: num_iso_H[0]=-1 means INCHI
|
110
|
+
adds implicit H automatically),
|
111
|
+
[1]: number of implicit isotopic 1H (protium),
|
112
|
+
[2]: number of implicit 2H (deuterium),
|
113
|
+
[3]: number of implicit 3H (tritium) */
|
114
|
+
AT_NUM isotopic_mass; /* 0 => non-isotopic; isotopic mass or */
|
115
|
+
/* ISOTOPIC_SHIFT_FLAG + mass - (average atomic mass) */
|
116
|
+
S_CHAR radical; /* inchi_Radical */
|
117
|
+
S_CHAR charge; /* positive or negative; 0 => no charge */
|
118
|
+
}inchi_Atom;
|
119
|
+
|
120
|
+
/*******************************************************************
|
121
|
+
* Notes: 1. Atom ordering numbers (i, k, and atom[i].neighbor[j] below)
|
122
|
+
* start from zero; max. ordering number is (num_atoms-1).
|
123
|
+
* 2. inchi_Atom atom[i] is connected to the atom[atom[i].neighbor[j]]
|
124
|
+
* by a bond that has type atom[i].bond_type[j] and 2D stereo type
|
125
|
+
* atom[i].bond_stereo[j] (in case of no stereo
|
126
|
+
* atom[i].bond_stereo[j] = INCHI_BOND_STEREO_NONE)
|
127
|
+
* Index j is in the range 0 <= j <= (atom[i].num_bonds-1)
|
128
|
+
* 3. Any connection (represented by atom[i].neighbor[j],
|
129
|
+
* atom[i].bond_type[j], and atom[i].bond_stereo[j])
|
130
|
+
* should be present in one or both adjacency list:
|
131
|
+
* if k = atom[i].neighbor[j] then i may or may not be present in
|
132
|
+
* atom[k].neighbor[] list. For example, the adjacency lists may be
|
133
|
+
* populated with only such neighbors that atom[i].neighbor[j] < i
|
134
|
+
* All elements of an adjacency list must be different, that is,
|
135
|
+
* a bond must be specified in an adjacency list only once.
|
136
|
+
* 4. in Molfiles usually
|
137
|
+
* (number of implicit H) = Valence - SUM(bond_type[])
|
138
|
+
* 5. Seemingly illogical order of the inchi_Atom members was
|
139
|
+
* chosen in an attempt to avoid alignment problems when
|
140
|
+
* accessing inchi_Atom from unrelated to C programming
|
141
|
+
* languages such as Visual Basic.
|
142
|
+
*******************************************************************/
|
143
|
+
|
144
|
+
/*******************************************************************
|
145
|
+
0D Stereo Parity and Type definitions
|
146
|
+
*******************************************************************
|
147
|
+
Note:
|
148
|
+
=====
|
149
|
+
o Below #A is the ordering number of atom A, starting from 0
|
150
|
+
o See parity values corresponding to 'o', 'e', and 'u' in
|
151
|
+
inchi_StereoParity0D definition below)
|
152
|
+
|
153
|
+
=============================================
|
154
|
+
stereogenic bond >A=B< or cumulene >A=C=C=B<
|
155
|
+
=============================================
|
156
|
+
|
157
|
+
neighbor[4] : {#X,#A,#B,#Y} in this order
|
158
|
+
X central_atom : NO_ATOM
|
159
|
+
\ X Y type : INCHI_StereoType_DoubleBond
|
160
|
+
A==B \ /
|
161
|
+
\ A==B
|
162
|
+
Y
|
163
|
+
|
164
|
+
parity= 'e' parity= 'o' unknown parity = 'u'
|
165
|
+
|
166
|
+
Limitations:
|
167
|
+
============
|
168
|
+
o Atoms A and B in cumulenes MUST be connected by a chain of double bonds;
|
169
|
+
atoms A and B in a stereogenic 'double bond' may be connected by a double,
|
170
|
+
single, or alternating bond.
|
171
|
+
o One atom may belong to up to 3 stereogenic bonds (i.g. in a fused
|
172
|
+
aromatic structure).
|
173
|
+
o Multiple stereogenic bonds incident to any given atom should
|
174
|
+
either all except possibly one have (possibly different) defined
|
175
|
+
parities ('o' or 'e') or should all have an unknown parity 'u'.
|
176
|
+
|
177
|
+
Note on parities of alternating stereobonds
|
178
|
+
===========================================
|
179
|
+
D--E
|
180
|
+
In large rings (see Fig. 1, all // \\
|
181
|
+
atoms are C) all alternating bonds B--C F--G
|
182
|
+
are treated as stereogenic. // \\
|
183
|
+
To avoid "undefined" bond parities A H
|
184
|
+
for bonds BC, DE, FG, HI, JK, LM, AN \ /
|
185
|
+
it is recommended to mark them with N==M J==I
|
186
|
+
parities. \ /
|
187
|
+
L==K Fig. 1
|
188
|
+
Such a marking will make
|
189
|
+
the stereochemical layer unambiguous
|
190
|
+
and it will be different from the B--C F--G
|
191
|
+
stereochemical layer of the second // \\ // \\
|
192
|
+
structure (Fig. 2). A D--E H
|
193
|
+
\ /
|
194
|
+
N==M J==I
|
195
|
+
By default, double and alternating \ /
|
196
|
+
bonds in 8-member and greater rings L==K Fig. 2
|
197
|
+
are treated by InChI as stereogenic.
|
198
|
+
|
199
|
+
|
200
|
+
=============================================
|
201
|
+
tetrahedral atom
|
202
|
+
=============================================
|
203
|
+
|
204
|
+
4 neighbors
|
205
|
+
|
206
|
+
X neighbor[4] : {#W, #X, #Y, #Z}
|
207
|
+
| central_atom: #A
|
208
|
+
W--A--Y type : INCHI_StereoType_Tetrahedral
|
209
|
+
|
|
210
|
+
Z
|
211
|
+
parity: if (X,Y,Z) are clockwize when seen from W then parity is 'e' otherwise 'o'
|
212
|
+
Example (see AXYZW above): if W is above the plane XYZ then parity = 'e'
|
213
|
+
|
214
|
+
3 neighbors
|
215
|
+
|
216
|
+
Y Y neighbor[4] : {#A, #X, #Y, #Z}
|
217
|
+
/ / central_atom: #A
|
218
|
+
X--A (e.g. O=S ) type : INCHI_StereoType_Tetrahedral
|
219
|
+
\ \
|
220
|
+
Z Z
|
221
|
+
|
222
|
+
parity: if (X,Y,Z) are clockwize when seen from A then parity is 'e',
|
223
|
+
otherwise 'o'
|
224
|
+
unknown parity = 'u'
|
225
|
+
Example (see AXYZ above): if A is above the plane XYZ then parity = 'e'
|
226
|
+
This approach may be used also in case of an implicit H attached to A.
|
227
|
+
|
228
|
+
=============================================
|
229
|
+
allene
|
230
|
+
=============================================
|
231
|
+
|
232
|
+
X Y neighbor[4] : {#X,#A,#B,#Y}
|
233
|
+
\ / central_atom : #C
|
234
|
+
A=C=B type : INCHI_StereoType_Allene
|
235
|
+
|
236
|
+
Y X
|
237
|
+
| |
|
238
|
+
when seen from A along A=C=B: X-A Y-A
|
239
|
+
|
240
|
+
parity: 'e' 'o'
|
241
|
+
|
242
|
+
parity: if A, B, Y are clockwise when seen from X then parity is 'e',
|
243
|
+
otherwise 'o'
|
244
|
+
unknown parity = 'u'
|
245
|
+
Example (see XACBY above): if X on the diagram is above the plane ABY
|
246
|
+
then parity is 'o'
|
247
|
+
|
248
|
+
Limitations
|
249
|
+
===========
|
250
|
+
o Atoms A and B in allenes MUST be connected by a chain of double bonds;
|
251
|
+
|
252
|
+
==============================================
|
253
|
+
Note. Correspondence to CML 0D stereo parities
|
254
|
+
==============================================
|
255
|
+
a list of 4 atoms corresponds to CML atomRefs4
|
256
|
+
|
257
|
+
tetrahedral atom
|
258
|
+
================
|
259
|
+
CML atomParity > 0 <=> INCHI_PARITY_EVEN
|
260
|
+
CML atomParity < 0 <=> INCHI_PARITY_ODD
|
261
|
+
|
262
|
+
| 1 1 1 1 | where xW is x-coordinate of
|
263
|
+
| xW xX xY xZ | atom W, etc. (xyz is a
|
264
|
+
CML atomParity = determinant | yW yX yY yZ | 'right-handed' Cartesian
|
265
|
+
| zW zX xY zZ | coordinate system)
|
266
|
+
|
267
|
+
allene (not yet defined in CML)
|
268
|
+
===============================
|
269
|
+
the parity corresponds to the sign of the following determinant
|
270
|
+
in exactly same way as for tetrahedral atoms:
|
271
|
+
|
272
|
+
| 1 1 1 1 | where bonds and neighbor[4] array are
|
273
|
+
| xX xA xB xY | same as defined above for allenes
|
274
|
+
| yX yA yB yY | Obviously, the parity is same for
|
275
|
+
| zX zA xB zY | {#X,#A,#B,#Y} and {#Y,#B,#A,#X}
|
276
|
+
because of the even number of column permutations.
|
277
|
+
|
278
|
+
stereogenic double bond and (not yet defined in CML) cumulenes
|
279
|
+
==============================================================
|
280
|
+
CML 'C' (cis) <=> INCHI_PARITY_ODD
|
281
|
+
CML 'T' (trans) <=> INCHI_PARITY_EVEN
|
282
|
+
|
283
|
+
|
284
|
+
How InChI uses 0D parities
|
285
|
+
==========================
|
286
|
+
|
287
|
+
1. 0D parities are used if all atom coordinates are zeroes.
|
288
|
+
|
289
|
+
In addition to that:
|
290
|
+
|
291
|
+
2. 0D parities are used for Stereobonds, Allenes, or Cumulenes if:
|
292
|
+
|
293
|
+
2a. A bond to the end-atom is shorter than MIN_BOND_LEN=0.000001
|
294
|
+
2b. A ratio of two bond lengths to the end-atom is smaller than MIN_SINE=0.03
|
295
|
+
2c. In case of a linear fragment X-A=B end-atom A is treated as satisfying 2a-b
|
296
|
+
|
297
|
+
0D parities are used if 2a or 2b or 2c applies to one or both end-atoms.
|
298
|
+
|
299
|
+
3. 0D parities are used for Tetrahedral Atoms if at least one of 3a-c is true:
|
300
|
+
|
301
|
+
3a. One of bonds to the central atom is shorter than MIN_BOND_LEN=0.000001
|
302
|
+
3b. A ratio of two bond lengths to the central atom is smaller than MIN_SINE=0.03
|
303
|
+
3c. The four neighbors are almost in one plane or the central atom and
|
304
|
+
its only 3 explicit neighbors are almost in one plane
|
305
|
+
|
306
|
+
Notes on 0D parities and 'undefined' stereogenic elements
|
307
|
+
=========================================================
|
308
|
+
|
309
|
+
If 0D parity is to be used according to 1-3 but CH3 CH3
|
310
|
+
has not been provided then the corresponding \ /
|
311
|
+
stereogenic element is considered 'undefined'. C=CH
|
312
|
+
/
|
313
|
+
For example, if in the structure (Fig. 3) H
|
314
|
+
the explicit H has been moved so that it Fig. 3
|
315
|
+
has same coordinates as atom >C= (that is,
|
316
|
+
the length of the bond H-C became zero)
|
317
|
+
then the double bond is assigned 'undefined' CH3 CH3
|
318
|
+
parity which by default is omitted from the \ /
|
319
|
+
Identifier. CH=CH
|
320
|
+
|
321
|
+
However, the structure on Fig. 4 will have double Fig. 4
|
322
|
+
bond parity 'o' and its parity in the Identifier is (-).
|
323
|
+
|
324
|
+
Notes on 0D parities in structures containing metals
|
325
|
+
====================================================
|
326
|
+
Since InChI disconnects bonds to metals the 0D parities upon the
|
327
|
+
disconnection may change in several different ways:
|
328
|
+
|
329
|
+
1) previously non-stereogenic bond may become stereogenic:
|
330
|
+
|
331
|
+
\ / \ /
|
332
|
+
CH==CH disconnection CH==CH
|
333
|
+
\ / ======>
|
334
|
+
M M
|
335
|
+
|
336
|
+
before the disconnection: after the disconnection:
|
337
|
+
atoms C have valence=5 and the double bond may become
|
338
|
+
the double bond is not stereogenic
|
339
|
+
recognized as stereogenic
|
340
|
+
|
341
|
+
2) previously stereogenic bond may become non-stereogenic:
|
342
|
+
|
343
|
+
M M(+)
|
344
|
+
\ / /
|
345
|
+
N==C disconnection (-)N==C
|
346
|
+
\ ======> \
|
347
|
+
|
348
|
+
3) Oddball structures, usually resulting from projecting 3D
|
349
|
+
structures on the plane, may contain fragment like that
|
350
|
+
depicted on Fig. 5:
|
351
|
+
|
352
|
+
M A M A
|
353
|
+
|\ / B / B
|
354
|
+
| X / disconnection / /
|
355
|
+
|/ \ / ======> / /
|
356
|
+
C===C C===C
|
357
|
+
Fig. 5
|
358
|
+
(X stands for bond intersection)
|
359
|
+
|
360
|
+
A-C=C-B parity is A-C=C-B parity is
|
361
|
+
trans (e) cis (o) or undefined
|
362
|
+
because the bond because C valence = 3,
|
363
|
+
orientation is same not 4.
|
364
|
+
as on Fig, 6 below:
|
365
|
+
|
366
|
+
A M
|
367
|
+
\ / Removal of M from the structure
|
368
|
+
C===C on Fig. 5 changes the geometry from trans
|
369
|
+
/ \ to cis.
|
370
|
+
M' B Removal of M and M' from the structure
|
371
|
+
Fig. 6 on Fig. 6 does not change the A-C=C-B
|
372
|
+
geometry: it is trans.
|
373
|
+
|
374
|
+
To resolve the problem InChI API accepts the second parity
|
375
|
+
corresponding to the metal-disconnected structure.
|
376
|
+
To store both bond parities use left shift by 3 bits:
|
377
|
+
|
378
|
+
inchi_Stereo0D::parity = ParityOfConnected | (ParityOfDisconnected<<3)
|
379
|
+
|
380
|
+
In case when only disconnected structure parity exists set
|
381
|
+
ParityOfConnected = INCHI_PARITY_UNDEFINED.
|
382
|
+
This is the only case when INCHI_PARITY_UNDEFINED parity
|
383
|
+
may be fed to the InChI.
|
384
|
+
|
385
|
+
In cases when the bond parity in a disconnected structure exists and
|
386
|
+
differs from the parity in the connected structure the atoms A and B
|
387
|
+
should be non-metals.
|
388
|
+
|
389
|
+
****************************************************************************/
|
390
|
+
|
391
|
+
#define NO_ATOM (-1) /* non-existent (central) atom */
|
392
|
+
|
393
|
+
/* 0D parity types */
|
394
|
+
typedef enum tagINCHIStereoType0D {
|
395
|
+
INCHI_StereoType_None = 0,
|
396
|
+
INCHI_StereoType_DoubleBond = 1,
|
397
|
+
INCHI_StereoType_Tetrahedral = 2,
|
398
|
+
INCHI_StereoType_Allene = 3
|
399
|
+
} inchi_StereoType0D;
|
400
|
+
|
401
|
+
/* 0D parities */
|
402
|
+
typedef enum tagINCHIStereoParity0D {
|
403
|
+
INCHI_PARITY_NONE = 0,
|
404
|
+
INCHI_PARITY_ODD = 1, /* 'o' */
|
405
|
+
INCHI_PARITY_EVEN = 2, /* 'e' */
|
406
|
+
INCHI_PARITY_UNKNOWN = 3, /* 'u' */
|
407
|
+
INCHI_PARITY_UNDEFINED = 4 /* '?' -- should not be used; however, see Note above */
|
408
|
+
} inchi_StereoParity0D;
|
409
|
+
|
410
|
+
|
411
|
+
/*************************************************
|
412
|
+
*
|
413
|
+
*
|
414
|
+
* 0D - S T E R E O (if no coordinates given)
|
415
|
+
*
|
416
|
+
*
|
417
|
+
*************************************************/
|
418
|
+
|
419
|
+
|
420
|
+
typedef struct tagINCHIStereo0D {
|
421
|
+
AT_NUM neighbor[4]; /* 4 atoms always */
|
422
|
+
AT_NUM central_atom; /* central tetrahedral atom or a central */
|
423
|
+
/* atom of allene; otherwise NO_ATOM */
|
424
|
+
S_CHAR type; /* inchi_StereoType0D */
|
425
|
+
S_CHAR parity; /* inchi_StereoParity0D: may be a combination of two parities: */
|
426
|
+
/* ParityOfConnected | (ParityOfDisconnected << 3), see Note above */
|
427
|
+
}inchi_Stereo0D;
|
428
|
+
|
429
|
+
/*************************************************
|
430
|
+
*
|
431
|
+
*
|
432
|
+
* I N C h I D L L I n p u t
|
433
|
+
*
|
434
|
+
*
|
435
|
+
*************************************************/
|
436
|
+
|
437
|
+
typedef struct tagINCHI_Input {
|
438
|
+
/* the caller is responsible for the data allocation and deallocation */
|
439
|
+
inchi_Atom *atom; /* array of num_atoms elements */
|
440
|
+
inchi_Stereo0D *stereo0D; /* array of num_stereo0D 0D stereo elements or NULL */
|
441
|
+
char *szOptions; /* InChI options: space-delimited; each is preceded by */
|
442
|
+
/* '/' or '-' depending on OS and compiler */
|
443
|
+
AT_NUM num_atoms; /* number of atoms in the structure < 1024 */
|
444
|
+
AT_NUM num_stereo0D; /* number of 0D stereo elements */
|
445
|
+
}inchi_Input;
|
446
|
+
|
447
|
+
/*************************************************
|
448
|
+
*
|
449
|
+
*
|
450
|
+
* I N C h I D L L O u t p u t
|
451
|
+
*
|
452
|
+
*
|
453
|
+
*************************************************/
|
454
|
+
|
455
|
+
typedef struct tagINCHI_Output {
|
456
|
+
/* zero-terminated C-strings allocated by GetINCHI() */
|
457
|
+
/* to deallocate all of them call FreeINCHI() (see below) */
|
458
|
+
char *szInChI; /* InChI ASCIIZ string */
|
459
|
+
char *szAuxInfo; /* Aux info ASCIIZ string */
|
460
|
+
char *szMessage; /* Error/warning ASCIIZ message */
|
461
|
+
char *szLog; /* log-file ASCIIZ string, contains a human-readable list */
|
462
|
+
/* of recognized options and possibly an Error/warning message */
|
463
|
+
} inchi_Output;
|
464
|
+
|
465
|
+
/*************************************************
|
466
|
+
*
|
467
|
+
*
|
468
|
+
* I N C h I D L L I n t e r f a c e
|
469
|
+
*
|
470
|
+
*
|
471
|
+
*************************************************/
|
472
|
+
|
473
|
+
#if (defined( _WIN32 ) && defined( _MSC_VER ) && defined(INCHI_LINK_AS_DLL) )
|
474
|
+
/* Win32 & MS VC ++, compile and link as a DLL */
|
475
|
+
#ifdef _USRDLL
|
476
|
+
/* InChI library dll */
|
477
|
+
#define INCHI_API __declspec(dllexport)
|
478
|
+
#define EXPIMP_TEMPLATE
|
479
|
+
#define INCHI_DECL __stdcall
|
480
|
+
#else
|
481
|
+
/* calling the InChI dll program */
|
482
|
+
#define INCHI_API __declspec(dllimport)
|
483
|
+
#define EXPIMP_TEMPLATE extern
|
484
|
+
#define INCHI_DECL __stdcall
|
485
|
+
#endif
|
486
|
+
#else
|
487
|
+
/* create a statically linked InChI library or link to an executable */
|
488
|
+
#define INCHI_API
|
489
|
+
#define EXPIMP_TEMPLATE
|
490
|
+
#define INCHI_DECL
|
491
|
+
#endif
|
492
|
+
|
493
|
+
/* GetINCHI(...) and Get_inchi_Input_FromAuxInfo(...) return values: */
|
494
|
+
|
495
|
+
typedef enum tagRetValGetINCHI {
|
496
|
+
|
497
|
+
inchi_Ret_SKIP = -2, /* not used in InChI library */
|
498
|
+
inchi_Ret_EOF = -1, /* no structural data has been provided */
|
499
|
+
inchi_Ret_OKAY = 0, /* Success; no errors or warnings */
|
500
|
+
inchi_Ret_WARNING = 1, /* Success; warning(s) issued */
|
501
|
+
inchi_Ret_ERROR = 2, /* Error: no InChI has been created */
|
502
|
+
inchi_Ret_FATAL = 3, /* Severe error: no InChI has been created (typically, memory allocation failure) */
|
503
|
+
inchi_Ret_UNKNOWN = 4, /* Unknown program error */
|
504
|
+
inchi_Ret_BUSY = 5 /* Previuos call to InChI has not returned yet */
|
505
|
+
|
506
|
+
} RetValGetINCHI;
|
507
|
+
|
508
|
+
/* to compile all InChI code as a C++ code #define INCHI_ALL_CPP */
|
509
|
+
#ifndef INCHI_ALL_CPP
|
510
|
+
#ifdef __cplusplus
|
511
|
+
extern "C" {
|
512
|
+
#endif
|
513
|
+
#endif
|
514
|
+
|
515
|
+
/* inchi_Input is created by the user; strings in inchi_Output are allocated and deallocated by InChI */
|
516
|
+
/* inchi_Output does not need to be initilized out to zeroes; see FreeINCHI on how to deallocate it */
|
517
|
+
EXPIMP_TEMPLATE INCHI_API int INCHI_DECL GetINCHI( inchi_Input *inp, inchi_Output *out );
|
518
|
+
|
519
|
+
/* FreeINCHI() should be called to deallocate char* pointers obtained from each GetINCHI() call */
|
520
|
+
EXPIMP_TEMPLATE INCHI_API void INCHI_DECL FreeINCHI ( inchi_Output *out );
|
521
|
+
|
522
|
+
/* helper: get string length */
|
523
|
+
EXPIMP_TEMPLATE INCHI_API int INCHI_DECL GetStringLength( char *p );
|
524
|
+
|
525
|
+
|
526
|
+
|
527
|
+
#ifndef INCHI_ALL_CPP
|
528
|
+
#ifdef __cplusplus
|
529
|
+
}
|
530
|
+
#endif
|
531
|
+
#endif
|
532
|
+
|
533
|
+
/**********************************************************
|
534
|
+
Using inchi_Input::szOptions related to ChiralFlag
|
535
|
+
**********************************************************
|
536
|
+
Awailable options (use - instead of / for O.S. other than MS Windows):
|
537
|
+
|
538
|
+
/SUCF Use Chiral Flag
|
539
|
+
/ChiralFlagON Set Chiral Flag
|
540
|
+
/ChiralFlagOFF Set Not-Chiral Flag
|
541
|
+
|
542
|
+
The following table describes the effect of the options
|
543
|
+
---------------------------------------------------------------
|
544
|
+
Options Equivalent to Chiral Flag Information
|
545
|
+
stored in AuxInfo
|
546
|
+
---------------------------------------------------------------
|
547
|
+
/SUCF /ChiralFlagON /SABS Chiral Flag
|
548
|
+
/SUCF /ChiralFlagOFF /SREL Not-Chiral Flag
|
549
|
+
/SUCF /SREL none
|
550
|
+
/ChiralFlagON none Chiral Flag
|
551
|
+
/ChiralFlagOFF none Not-Chiral Flag
|
552
|
+
|
553
|
+
*************************************************************
|
554
|
+
Other options available in InChI Library (case insensitive)
|
555
|
+
*************************************************************
|
556
|
+
/SNon Exclude stereo (Default: Include Absolute stereo)
|
557
|
+
/SRel Relative stereo
|
558
|
+
/SRac Racemic stereo
|
559
|
+
|
560
|
+
/SUU Include omitted unknown/undefined stereo
|
561
|
+
/NEWPS Narrow end of wedge points to stereocenter (default: both)
|
562
|
+
/RecMet Include reconnected bond to metal results
|
563
|
+
/FixedH Mobile H Perception Off (Default: On)
|
564
|
+
/AuxNone Omit auxiliary information (default: Include)
|
565
|
+
/NoADP Disable Aggressive Deprotonation (for testing only)
|
566
|
+
/Compress Compressed output
|
567
|
+
/DoNotAddH Overrides inchi_Atom::num_iso_H[0] == -1
|
568
|
+
/Wnumber Set time-out per structure in seconds; W0 means unlimited
|
569
|
+
In InChI library the default value is unlimited
|
570
|
+
/OutputSDF Output SDfile instead of InChI
|
571
|
+
/WarnOnEmptyStructure Warn and produce empty InChI for empty structure
|
572
|
+
|
573
|
+
Empty InChI format: InChI=1//
|
574
|
+
AuxInfo=1//
|
575
|
+
*/
|
576
|
+
|
577
|
+
|
578
|
+
/*****************************************************************
|
579
|
+
*
|
580
|
+
*
|
581
|
+
* C o n v e r s i o n: InChI AuxInfo string => inchi_Input
|
582
|
+
*
|
583
|
+
*
|
584
|
+
*****************************************************************/
|
585
|
+
|
586
|
+
#ifndef STR_ERR_LEN
|
587
|
+
#define STR_ERR_LEN 256
|
588
|
+
#endif
|
589
|
+
|
590
|
+
typedef struct tagInchiInpData {
|
591
|
+
inchi_Input *pInp; /* a pointer to pInp that has all items 0 or NULL */
|
592
|
+
int bChiral; /* 1 => the structure was marked as chiral, 2=> not chiral, 0=> not marked */
|
593
|
+
char szErrMsg[STR_ERR_LEN];
|
594
|
+
} InchiInpData;
|
595
|
+
|
596
|
+
/* to compile all InChI code as a C++ code #define INCHI_ALL_CPP */
|
597
|
+
#ifndef INCHI_ALL_CPP
|
598
|
+
#ifdef __cplusplus
|
599
|
+
extern "C" {
|
600
|
+
#endif
|
601
|
+
#endif
|
602
|
+
|
603
|
+
/* Input
|
604
|
+
-----
|
605
|
+
szInchiAuxInfo: contains ASCIIZ string of InChI output for a single
|
606
|
+
structure or only the AuxInfo line
|
607
|
+
bDoNotAddH: if 0 then InChI will be allowed to add implicit H
|
608
|
+
pInchiInp: should have a valid pointer pInchiInp->pInp to an empty
|
609
|
+
(all members = 0) inchi_Input structure
|
610
|
+
|
611
|
+
Output
|
612
|
+
------
|
613
|
+
pInchiInp: The following members of pInp may be filled during the call:
|
614
|
+
atom, num_atoms, stereo0D, num_stereo0D
|
615
|
+
Return value: see RetValGetINCHI
|
616
|
+
*/
|
617
|
+
EXPIMP_TEMPLATE INCHI_API int INCHI_DECL Get_inchi_Input_FromAuxInfo
|
618
|
+
( char *szInchiAuxInfo, int bDoNotAddH, InchiInpData *pInchiInp );
|
619
|
+
|
620
|
+
/*
|
621
|
+
To deallocate and write zeroes into the changed members of pInchiInp->pInp call
|
622
|
+
Free_inchi_Input( inchi_Input *pInp )
|
623
|
+
*/
|
624
|
+
EXPIMP_TEMPLATE INCHI_API void INCHI_DECL Free_inchi_Input( inchi_Input *pInp );
|
625
|
+
|
626
|
+
#ifndef INCHI_ALL_CPP
|
627
|
+
#ifdef __cplusplus
|
628
|
+
}
|
629
|
+
#endif
|
630
|
+
#endif
|
631
|
+
|
632
|
+
|
633
|
+
/* Annotated List of -Exports for libinchi.dll produced by dumpbin (Win32):
|
634
|
+
|
635
|
+
ordinal hint RVA name
|
636
|
+
--------------------------------- C calling conventions, compatible with gcc under Win32
|
637
|
+
(see inchi_dll.c and vc6_libinchi.def for details)
|
638
|
+
1 0 0004D800 FreeINCHI
|
639
|
+
2 1 0004D840 Free_inchi_Input
|
640
|
+
3 2 0004D7F0 GetINCHI
|
641
|
+
4 3 0004D810 GetStringLength
|
642
|
+
5 4 0004D820 Get_inchi_Input_FromAuxInfo
|
643
|
+
--------------------------------- __stdcall calling conventions, compatible with MS VB 6
|
644
|
+
(as defined in this file and used in INCHI_MAIN.exe)
|
645
|
+
6 5 0004BD30 _FreeINCHI@4
|
646
|
+
7 6 0001D7C0 _Free_inchi_Input@4
|
647
|
+
8 7 0004BD80 _GetINCHI@8
|
648
|
+
9 8 0004D7D0 _GetStringLength@4
|
649
|
+
10 9 0001D620 _Get_inchi_Input_FromAuxInfo@12
|
650
|
+
|
651
|
+
------------------- end of the Annotated list-----------------------
|
652
|
+
|
653
|
+
=======================================================================
|
654
|
+
============= prototypes for C calling conventions: ===================
|
655
|
+
=======================================================================
|
656
|
+
int GetINCHI( inchi_Input *inp, inchi_Output *out );
|
657
|
+
void FreeINCHI( inchi_Output *out );
|
658
|
+
int GetStringLength( char *p );
|
659
|
+
int Get_inchi_Input_FromAuxInfo
|
660
|
+
( char *szInchiAuxInfo, int bDoNotAddH, InchiInpData *pInchiInp );
|
661
|
+
void Free_inchi_Input( inchi_Input *pInp );
|
662
|
+
=======================================================================
|
663
|
+
|
664
|
+
|
665
|
+
|
666
|
+
*/
|
667
|
+
|
668
|
+
/* Currently there is no callback function for aborting, progress, etc. */
|
669
|
+
|
670
|
+
#endif /* __INHCH_API_H__ */
|