/usr/include/rdkit/GraphMol/MolOps.h is in librdkit-dev 201503-3.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 | //
// Copyright (C) 2001-2012 Greg Landrum and Rational Discovery LLC
// Copyright (c) 2014, Novartis Institutes for BioMedical Research Inc.
//
// @@ All Rights Reserved @@
// This file is part of the RDKit.
// The contents are covered by the terms of the BSD license
// which is included in the file license.txt, found at the root
// of the RDKit source tree.
//
#ifndef _RD_MOL_OPS_H_
#define _RD_MOL_OPS_H_
#include <vector>
#include <map>
#include <list>
#include <boost/smart_ptr.hpp>
#include <boost/dynamic_bitset.hpp>
extern const int ci_LOCAL_INF;
namespace RDKit{
class ROMol;
class RWMol;
class Atom;
class Bond;
typedef std::vector<double> INVAR_VECT;
typedef INVAR_VECT::iterator INVAR_VECT_I;
typedef INVAR_VECT::const_iterator INVAR_VECT_CI;
//! \brief Groups a variety of molecular query and transformation operations.
namespace MolOps {
//! return the number of electrons available on an atom to donate for aromaticity
/*!
The result is determined using the default valency, number of lone pairs,
number of bonds and the formal charge. Note that the atom may not donate
all of these electrons to a ring for aromaticity (also used in Conjugation
and hybridization code).
\param at the atom of interest
\return the number of electrons
*/
int countAtomElec(const Atom *at);
//! sums up all atomic formal charges and returns the result
int getFormalCharge(const ROMol &mol);
//! returns whether or not the given Atom is involved in a conjugated bond
bool atomHasConjugatedBond(const Atom *at);
//! find fragments (disconnected components of the molecular graph)
/*!
\param mol the molecule of interest
\param mapping used to return the mapping of Atoms->fragments.
On return \c mapping will be <tt>mol->getNumAtoms()</tt> long
and will contain the fragment assignment for each Atom
\return the number of fragments found.
*/
unsigned int getMolFrags(const ROMol &mol,std::vector<int> &mapping);
//! find fragments (disconnected components of the molecular graph)
/*!
\param mol the molecule of interest
\param frags used to return the Atoms in each fragment
On return \c mapping will be \c numFrags long, and each entry
will contain the indices of the Atoms in that fragment.
\return the number of fragments found.
*/
unsigned int getMolFrags(const ROMol &mol, std::vector<std::vector<int> > &frags);
//! splits a molecule into its component fragments
// (disconnected components of the molecular graph)
/*!
\param mol the molecule of interest
\param sanitizeFrags toggles sanitization of the fragments after
they are built
\param frags used to return the mapping of Atoms->fragments.
if provided, \c frags will be <tt>mol->getNumAtoms()</tt> long
on return and will contain the fragment assignment for each Atom
\param fragsMolAtomMapping used to return the Atoms in each fragment
On return \c mapping will be \c numFrags long, and each entry
will contain the indices of the Atoms in that fragment.
\param copyConformers toggles copying conformers of the fragments after
they are built
\return a vector of the fragments as smart pointers to ROMols
*/
std::vector<boost::shared_ptr<ROMol> > getMolFrags(const ROMol &mol,
bool sanitizeFrags=true,
std::vector<int> *frags=0,
std::vector<std::vector<int> > *fragsMolAtomMapping=0,
bool copyConformers=true);
//! splits a molecule into pieces based on labels assigned using a query
/*!
\param mol the molecule of interest
\param query the query used to "label" the molecule for fragmentation
\param sanitizeFrags toggles sanitization of the fragments after
they are built
\param whiteList if provided, only labels in the list will be kept
\param negateList if true, the white list logic will be inverted: only labels
not in the list will be kept
\return a map of the fragments and their labels
*/
template <typename T>
std::map<T,boost::shared_ptr<ROMol> > getMolFragsWithQuery(const ROMol &mol,
T (*query)(const ROMol &,const Atom *),
bool sanitizeFrags=true,
const std::vector<T> *whiteList=0,
bool negateList=false);
#if 0
//! finds a molecule's minimium spanning tree (MST)
/*!
\param mol the molecule of interest
\param mst used to return the MST as a vector of bond indices
*/
void findSpanningTree(const ROMol &mol,std::vector<int> &mst);
#endif
//! calculates Balaban's J index for the molecule
/*!
\param mol the molecule of interest
\param useBO toggles inclusion of the bond order in the calculation
(when false, we're not really calculating the J value)
\param force forces the calculation (instead of using cached results)
\param bondPath when included, only paths using bonds whose indices occur
in this vector will be included in the calculation
\param cacheIt If this is true, the calculated value will be cached
as a property on the molecule
\return the J index
*/
double computeBalabanJ(const ROMol &mol,
bool useBO=true,
bool force=false,
const std::vector<int> *bondPath=0,
bool cacheIt=true);
//! \overload
double computeBalabanJ(double *distMat, int nb, int nAts);
//! \name Dealing with hydrogens
//{@
//! returns a copy of a molecule with hydrogens added in as explicit Atoms
/*!
\param mol the molecule to add Hs to
\param explicitOnly (optional) if this \c true, only explicit Hs will be added
\param addCoords (optional) If this is true, estimates for the atomic coordinates
of the added Hs will be used.
\return the new molecule
<b>Notes:</b>
- it makes no sense to use the \c addCoords option if the molecule's heavy
atoms don't already have coordinates.
- the caller is responsible for <tt>delete</tt>ing the pointer this returns.
*/
ROMol *addHs(const ROMol &mol,bool explicitOnly=false,bool addCoords=false);
//! \overload
// modifies the molecule in place
void addHs(RWMol &mol,bool explicitOnly=false,bool addCoords=false);
//! returns a copy of a molecule with hydrogens removed
/*!
\param mol the molecule to remove Hs from
\param implicitOnly (optional) if this \c true, only implicit Hs will be removed
\param updateExplicitCount (optional) If this is \c true, when explicit Hs are removed
from the graph, the heavy atom to which they are bound will have its counter of
explicit Hs increased.
\param sanitize: (optional) If this is \c true, the final molecule will be
sanitized
\return the new molecule
<b>Notes:</b>
- Hydrogens which aren't connected to a heavy atom will not be
removed. This prevents molecules like <tt>"[H][H]"</tt> from having
all atoms removed.
- Labelled hydrogen (e.g. atoms with atomic number=1, but mass > 1),
will not be removed.
- two coordinate Hs, like the central H in C[H-]C, will not be removed
- Hs connected to dummy atoms will not be removed
- the caller is responsible for <tt>delete</tt>ing the pointer this returns.
*/
ROMol *removeHs(const ROMol &mol,bool implicitOnly=false,
bool updateExplicitCount=false,bool sanitize=true);
//! \overload
// modifies the molecule in place
void removeHs(RWMol &mol,bool implicitOnly=false,
bool updateExplicitCount=false,bool sanitize=true);
//! returns a copy of a molecule with hydrogens removed and added as queries
//! to the heavy atoms to which they are bound.
/*!
This is really intended to be used with molecules that contain QueryAtoms
\param mol the molecule to remove Hs from
\return the new molecule
<b>Notes:</b>
- Atoms that do not already have hydrogen count queries will have one
added, other H-related queries will not be touched. Examples:
- C[H] -> [C;!H0]
- [C;H1][H] -> [C;H1]
- [C;H2][H] -> [C;H2]
- Hydrogens which aren't connected to a heavy atom will not be
removed. This prevents molecules like <tt>"[H][H]"</tt> from having
all atoms removed.
- the caller is responsible for <tt>delete</tt>ing the pointer this returns.
- By default all hydrogens are removed, however if
mergeUnmappedOnly is true, any hydrogen participating
in an atom map will be retained
*/
ROMol *mergeQueryHs(const ROMol &mol, bool mergeUnmappedOnly=false);
//! \overload
// modifies the molecule in place
void mergeQueryHs(RWMol &mol, bool mergeUnmappedOnly=false);
//! returns a copy of a molecule with the atoms renumbered
/*!
\param mol the molecule to work with
\param newOrder the new ordering of the atoms (should be numAtoms long)
for example: if newOrder is [3,2,0,1], then atom 3 in the original
molecule will be atom 0 in the new one
\return the new molecule
<b>Notes:</b>
- the caller is responsible for <tt>delete</tt>ing the pointer this returns.
*/
ROMol *renumberAtoms(const ROMol &mol,const std::vector<unsigned int> &newOrder);
//@}
//! \name Sanitization
//@{
typedef enum {
SANITIZE_NONE=0x0,
SANITIZE_CLEANUP=0x1,
SANITIZE_PROPERTIES=0x2,
SANITIZE_SYMMRINGS=0x4,
SANITIZE_KEKULIZE=0x8,
SANITIZE_FINDRADICALS=0x10,
SANITIZE_SETAROMATICITY=0x20,
SANITIZE_SETCONJUGATION=0x40,
SANITIZE_SETHYBRIDIZATION=0x80,
SANITIZE_CLEANUPCHIRALITY=0x100,
SANITIZE_ADJUSTHS=0x200,
SANITIZE_ALL=0xFFFFFFF
} SanitizeFlags;
//! \brief carries out a collection of tasks for cleaning up a molecule and ensuring
//! that it makes "chemical sense"
/*!
This functions calls the following in sequence
-# MolOps::cleanUp()
-# mol.updatePropertyCache()
-# MolOps::symmetrizeSSSR()
-# MolOps::Kekulize()
-# MolOps::assignRadicals()
-# MolOps::setAromaticity()
-# MolOps::setConjugation()
-# MolOps::setHybridization()
-# MolOps::cleanupChirality()
-# MolOps::adjustHs()
\param mol : the RWMol to be cleaned
\param operationThatFailed : the first (if any) sanitization operation that fails is set here.
The values are taken from the \c SanitizeFlags enum.
On success, the value is \c SanitizeFlags::SANITIZE_NONE
\param sanitizeOps : the bits here are used to set which sanitization operations are carried
out. The elements of the \c SanitizeFlags enum define the operations.
<b>Notes:</b>
- If there is a failure in the sanitization, a \c SanitException
will be thrown.
- in general the user of this function should cast the molecule following this
function to a ROMol, so that new atoms and bonds cannot be added to the
molecule and screw up the sanitizing that has been done here
*/
void sanitizeMol(RWMol &mol,unsigned int &operationThatFailed,
unsigned int sanitizeOps=SANITIZE_ALL);
//! \overload
void sanitizeMol(RWMol &mol);
//! Sets up the aromaticity for a molecule
/*!
This is what happens here:
-# find all the simple rings by calling the findSSSR function
-# loop over all the Atoms in each ring and mark them if they are candidates
for aromaticity. A ring atom is a candidate if it can spare electrons
to the ring and if it's from the first two rows of the periodic table.
-# ased on the candidate atoms, mark the rings to be either candidates
or non-candidates. A ring is a candidate only if all its atoms are candidates
-# apply Hueckel rule to each of the candidate rings to check if the ring can be
aromatic
\param mol the RWMol of interest
\return 1 on succes, 0 otherwise
<b>Assumptions:</b>
- Kekulization has been done (i.e. \c MolOps::Kekulize() has already
been called)
*/
int setAromaticity(RWMol &mol);
//! Designed to be called by the sanitizer to handle special cases before anything is done.
/*!
Currently this:
- modifies nitro groups, so that the nitrogen does not have a unreasonable
valence of 5, as follows:
- the nitrogen gets a positve charge
- one of the oxygens gets a negative chage and the double bond to this
oxygen is changed to a single bond
The net result is that nitro groups can be counted on to be:
\c "[N+](=O)[O-]"
\param mol the molecule of interest
*/
void cleanUp(RWMol &mol);
//! Called by the sanitizer to assign radical counts to atoms
void assignRadicals(RWMol &mol);
//! adjust the number of implicit and explicit Hs for special cases
/*!
Currently this:
- modifies aromatic nitrogens so that, when appropriate, they have an
explicit H marked (e.g. so that we get things like \c "c1cc[nH]cc1"
\param mol the molecule of interest
<b>Assumptions</b>
- this is called after the molecule has been sanitized,
aromaticity has been perceived, and the implicit valence of
everything has been calculated.
*/
void adjustHs(RWMol &mol);
//! Kekulizes the molecule
/*!
\param mol the molecule of interest
\param markAtomsBonds if this is set to true, \c isAromatic boolean settings
on both the Bonds and Atoms are turned to false following
the Kekulization, otherwise they are left alone in their
original state.
\param maxBackTracks the maximum number of attempts at back-tracking. The algorithm
uses a back-tracking procedure to revist a previous setting of
double bond if we hit a wall in the kekulization process
<b>Notes:</b>
- even if \c markAtomsBonds is \c false the \c BondType for all aromatic
bonds will be changed from \c RDKit::Bond::AROMATIC to \c RDKit::Bond::SINGLE
or RDKit::Bond::DOUBLE during Kekulization.
*/
void Kekulize(RWMol &mol, bool markAtomsBonds=true, unsigned int maxBackTracks=100);
//! flags the molecule's conjugated bonds
void setConjugation(ROMol &mol);
//! calculates and sets the hybridization of all a molecule's Stoms
void setHybridization(ROMol &mol);
// @}
//! \name Ring finding and SSSR
//@{
//! finds a molecule's Smallest Set of Smallest Rings
/*!
Currently this implements a modified form of Figueras algorithm
(JCICS - Vol. 36, No. 5, 1996, 986-991)
\param mol the molecule of interest
\param res used to return the vector of rings. Each entry is a vector with
atom indices. This information is also stored in the molecule's
RingInfo structure, so this argument is optional (see overload)
\return number of smallest rings found
Base algorithm:
- The original algorithm starts by finding representative degree 2
nodes.
- Representative because if a series of deg 2 nodes are found only
one of them is picked.
- The smallest ring around each of them is found.
- The bonds that connect to this degree 2 node are them chopped off, yielding
new deg two nodes
- The process is repeated on the new deg 2 nodes.
- If no deg 2 nodes are found, a deg 3 node is picked. The smallest ring
with it is found. A bond from this is "carefully" (look in the paper)
selected and chopped, yielding deg 2 nodes. The process is same as
above once this is done.
Our Modifications:
- If available, more than one smallest ring around a representative deg 2
node will be computed and stored
- Typically 3 rings are found around a degree 3 node (when no deg 2s are available)
and all the bond to that node are chopped.
- The extra rings that were found in this process are removed after all the nodes
have been covered.
These changes were motivated by several factors:
- We believe the original algorithm fails to find the correct SSSR
(finds the correct number of them but the wrong ones) on some sample mols
- Since SSSR may not be unique, a post-SSSR step to symmetrize may be done.
The extra rings this process adds can be quite useful.
*/
int findSSSR(const ROMol &mol, std::vector<std::vector<int> > &res);
//! \overload
int findSSSR(const ROMol &mol, std::vector<std::vector<int> > *res=0);
//! use a DFS algorithm to identify ring bonds and atoms in a molecule
/*!
\b NOTE: though the RingInfo structure is populated by this function,
the only really reliable calls that can be made are to check if
mol.getRingInfo().numAtomRings(idx) or mol.getRingInfo().numBondRings(idx)
return values >0
*/
void fastFindRings(const ROMol &mol);
//! symmetrize the molecule's Smallest Set of Smallest Rings
/*!
SSSR rings obatined from "findSSSR" can be non-unique in some case.
For example, cubane has five SSSR rings, not six as one would hope.
This function adds additional rings to the SSSR list if necessary
to make the list symmetric, e.g. all atoms in cubane will be part of the same number
of SSSRs. This function choses these extra rings from the extra rings computed
and discarded during findSSSR. The new ring are chosen such that:
- replacing a same sized ring in the SSSR list with an extra ring yields
the same union of bond IDs as the orignal SSSR list
\param mol - the molecule of interest
\param res used to return the vector of rings. Each entry is a vector with
atom indices. This information is also stored in the molecule's
RingInfo structure, so this argument is optional (see overload)
\return the total number of rings = (new rings + old SSSRs)
<b>Notes:</b>
- if no SSSR rings are found on the molecule - MolOps::findSSSR() is called first
*/
int symmetrizeSSSR(ROMol &mol, std::vector<std::vector<int> > &res);
//! \overload
int symmetrizeSSSR(ROMol &mol);
//@}
//! \name Shortest paths and other matrices
//@{
//! returns a molecule's adjacency matrix
/*!
\param mol the molecule of interest
\param useBO toggles use of bond orders in the matrix
\param emptyVal sets the empty value (for non-adjacent atoms)
\param force forces calculation of the matrix, even if already computed
\param propNamePrefix used to set the cached property name
\return the adjacency matrix.
<b>Notes</b>
- The result of this is cached in the molecule's local property dictionary,
which will handle deallocation. Do the caller should <b>not</b> \c delete
this pointer.
*/
double * getAdjacencyMatrix(const ROMol &mol,
bool useBO=false,
int emptyVal=0,
bool force=false,
const char *propNamePrefix=0,
const boost::dynamic_bitset<> *bondsToUse=0
);
//! Computes the molecule's topological distance matrix
/*!
Uses the Floyd-Warshall all-pairs-shortest-paths algorithm.
\param mol the molecule of interest
\param useBO toggles use of bond orders in the matrix
\param useAtomWts sets the diagonal elements of the result to
6.0/(atomic number) so that the matrix can be used to calculate
Balaban J values. This does not affect the bond weights.
\param force forces calculation of the matrix, even if already computed
\param propNamePrefix used to set the cached property name
\return the distance matrix.
<b>Notes</b>
- The result of this is cached in the molecule's local property dictionary,
which will handle deallocation. Do the caller should <b>not</b> \c delete
this pointer.
*/
double *getDistanceMat(const ROMol &mol,
bool useBO=false,
bool useAtomWts=false,
bool force=false,
const char *propNamePrefix=0);
//! Computes the molecule's topological distance matrix
/*!
Uses the Floyd-Warshall all-pairs-shortest-paths algorithm.
\param mol the molecule of interest
\param activeAtoms only elements corresponding to these atom indices
will be included in the calculation
\param bonds only bonds found in this list will be included in the
calculation
\param useBO toggles use of bond orders in the matrix
\param useAtomWts sets the diagonal elements of the result to
6.0/(atomic number) so that the matrix can be used to calculate
Balaban J values. This does not affect the bond weights.
\return the distance matrix.
<b>Notes</b>
- The results of this call are not cached, the caller <b>should</b> \c delete
this pointer.
*/
double *getDistanceMat(const ROMol &mol,
const std::vector<int> &activeAtoms,
const std::vector<const Bond *> &bonds,
bool useBO=false,
bool useAtomWts=false);
//! Computes the molecule's 3D distance matrix
/*!
\param mol the molecule of interest
\param confId the conformer to use
\param useAtomWts sets the diagonal elements of the result to
6.0/(atomic number)
\param force forces calculation of the matrix, even if already computed
\param propNamePrefix used to set the cached property name
\return the distance matrix.
<b>Notes</b>
- The result of this is cached in the molecule's local property dictionary,
which will handle deallocation. Do the caller should <b>not</b> \c delete
this pointer.
*/
double *get3DDistanceMat(const ROMol &mol,
int confId=-1,
bool useAtomWts=false,
bool force=false,
const char *propNamePrefix=0);
//! Find the shortest path between two atoms
/*!
Uses the Bellman-Ford algorithm
\param mol molecule of interest
\param aid1 index of the first atom
\param aid2 index of the second atom
\return an std::list with the indices of the atoms along the shortest
path
<b>Notes:</b>
- the starting and end atoms are included in the path
- if no path is found, an empty path is returned
*/
std::list<int> getShortestPath(const ROMol &mol, int aid1, int aid2);
//@}
#if 0
//! \name Canonicalization
//@{
//! assign a canonical ordering to a molecule's atoms
/*!
The algorithm used here is a modification of the published Daylight canonical
smiles algorithm (i.e. it uses atom invariants and products of primes).
\param mol the molecule of interest
\param ranks used to return the ranks
\param breakTies toggles breaking of ties (see below)
\param includeChirality toggles inclusion of chirality in the invariants
\param includeIsotopes toggles inclusion of isotopes in the invariants
\param rankHistory used to return the rank history (see below)
<b>Notes:</b>
- Tie breaking should be done when it's important to have a full ordering
of the atoms (e.g. when generating canonical traversal trees). If it's
acceptable to have ties between symmetry-equivalent atoms (e.g. when
generating CIP codes), tie breaking can/should be skipped.
- if the \c rankHistory argument is provided, the evolution of the ranks of
individual atoms will be tracked. The \c rankHistory pointer should be
to a VECT_INT_VECT that has at least \c mol.getNumAtoms() elements.
*/
void rankAtoms(const ROMol &mol,std::vector<int> &ranks,
bool breakTies=true,
bool includeChirality=true,
bool includeIsotopes=true,
std::vector<std::vector<int> > *rankHistory=0);
//! assign a canonical ordering to a sub-molecule's atoms
/*!
The algorithm used here is a modification of the published Daylight canonical
smiles algorithm (i.e. it uses atom invariants and products of primes).
\param mol the molecule of interest
\param atomsToUse atoms to be included
\param bondsToUse bonds to be included
\param atomSymbols symbols to use for the atoms in the output (these are
used in place of atomic number and isotope information)
\param ranks used to return the ranks
\param breakTies toggles breaking of ties (see below)
\param rankHistory used to return the rank history (see below)
<b>Notes:</b>
- Tie breaking should be done when it's important to have a full ordering
of the atoms (e.g. when generating canonical traversal trees). If it's
acceptable to have ties between symmetry-equivalent atoms (e.g. when
generating CIP codes), tie breaking can/should be skipped.
- if the \c rankHistory argument is provided, the evolution of the ranks of
individual atoms will be tracked. The \c rankHistory pointer should be
to a VECT_INT_VECT that has at least \c mol.getNumAtoms() elements.
*/
void rankAtomsInFragment(const ROMol &mol,std::vector<int> &ranks,
const boost::dynamic_bitset<> &atomsToUse,
const boost::dynamic_bitset<> &bondsToUse,
const std::vector<std::string> *atomSymbols=0,
const std::vector<std::string> *bondSymbols=0,
bool breakTies=true,
std::vector<std::vector<int> > *rankHistory=0);
// @}
#endif
//! \name Stereochemistry
//@{
//! removes bogus chirality markers (those on non-sp3 centers):
void cleanupChirality(RWMol &mol);
//! \brief Uses a conformer to assign ChiralType to a molecule's atoms
/*!
\param mol the molecule of interest
\param confId the conformer to use
\param replaceExistingTags if this flag is true, any existing atomic chiral
tags will be replaced
If the conformer provided is not a 3D conformer, nothing will be done.
*/
void assignChiralTypesFrom3D(ROMol &mol,int confId=-1,bool replaceExistingTags=true);
//! Assign stereochemistry tags to atoms (i.e. R/S) and bonds (i.e. Z/E)
/*!
\param mol the molecule of interest
\param cleanIt toggles removal of stereo flags from double bonds that can
not have stereochemistry
\param force forces the calculation to be repeated even if it has
already been done
\param flagPossibleStereoCenters set the _ChiralityPossible property on
atoms that are possible stereocenters
<b>Notes:M</b>
- Throughout we assume that we're working with a hydrogen-suppressed
graph.
*/
void assignStereochemistry(ROMol &mol,bool cleanIt=false,bool force=false,
bool flagPossibleStereoCenters=false);
//! Removes all stereochemistry information from atoms (i.e. R/S) and bonds (i.e. Z/E)
/*!
\param mol the molecule of interest
*/
void removeStereochemistry(ROMol &mol);
//! \brief finds bonds that could be cis/trans in a molecule and mark them as
//! Bond::STEREONONE
/*!
\param mol the molecule of interest
\param cleanIt toggles removal of stereo flags from double bonds that can
not have stereochemistry
This function is usefuly in two situations
- when parsing a mol file; for the bonds marked here, coordinate informations
on the neighbors can be used to indentify cis or trans states
- when writing a mol file; bonds that can be cis/trans but not marked as either
need to be specially marked in the mol file
*/
void findPotentialStereoBonds(ROMol &mol,bool cleanIt=false);
//@}
//! returns the number of atoms which have a particular property set
unsigned getNumAtomsWithDistinctProperty(const ROMol& mol, std::string prop);
}; // end of namespace MolOps
}; // end of namespace RDKit
#endif
|