RDKit
Open-source cheminformatics and machine learning.
RDDepict Namespace Reference

Classes

class  EmbeddedAtom
 Class that contains the data for an atoms that has alredy been embedded. More...
 
class  EmbeddedFrag
 Class containing a fragment of a molecule that has already been embedded. More...
 
struct  gtIIPair
 

Typedefs

typedef std::vector< const RDGeom::Point2D * > VECT_C_POINT
 
typedef std::pair< int, int > PAIR_I_I
 
typedef std::vector< PAIR_I_IVECT_PII
 
typedef std::priority_queue< PAIR_I_I, VECT_PII, gtIIPairPR_QUEUE
 
typedef std::pair< double, PAIR_I_IPAIR_D_I_I
 
typedef std::list< PAIR_D_I_ILIST_PAIR_DII
 
typedef std::pair< int, int > INT_PAIR
 
typedef std::vector< INT_PAIRINT_PAIR_VECT
 
typedef INT_PAIR_VECT::const_iterator INT_PAIR_VECT_CI
 
typedef std::pair< double, INT_PAIRDOUBLE_INT_PAIR
 
typedef boost::shared_array< double > DOUBLE_SMART_PTR
 
typedef std::map< unsigned int, EmbeddedAtomINT_EATOM_MAP
 
typedef INT_EATOM_MAP::iterator INT_EATOM_MAP_I
 
typedef INT_EATOM_MAP::const_iterator INT_EATOM_MAP_CI
 

Functions

void compute2DCoordsForReaction (RDKit::ChemicalReaction &rxn, double spacing=2.0, bool updateProps=true, bool canonOrient=false, unsigned int nFlipsPerSample=0, unsigned int nSamples=0, int sampleSeed=0, bool permuteDeg4Nodes=false)
 Generate 2D coordinates (a depiction) for a reaction. More...
 
RDGeom::INT_POINT2D_MAP embedRing (const RDKit::INT_VECT &ring)
 Some utility functions used in generating 2D coordinates. More...
 
void transformPoints (RDGeom::INT_POINT2D_MAP &nringCor, const RDGeom::Transform2D &trans)
 
RDGeom::Point2D computeBisectPoint (const RDGeom::Point2D &rcr, double ang, const RDGeom::Point2D &nb1, const RDGeom::Point2D &nb2)
 Find a point that bisects the angle at rcr. More...
 
void reflectPoints (RDGeom::INT_POINT2D_MAP &coordMap, const RDGeom::Point2D &loc1, const RDGeom::Point2D &loc2)
 Reflect a set of point through a the line joining two point. More...
 
RDGeom::Point2D reflectPoint (const RDGeom::Point2D &point, const RDGeom::Point2D &loc1, const RDGeom::Point2D &loc2)
 
RDKit::INT_VECT setNbrOrder (unsigned int aid, const RDKit::INT_VECT &nbrs, const RDKit::ROMol &mol)
 Set the neighbors yet to added to aid such that the atoms with the most subs. More...
 
RDKit::INT_VECT findNextRingToEmbed (const RDKit::INT_VECT &doneRings, const RDKit::VECT_INT_VECT &fusedRings, int &nextId)
 From a given set of rings find the ring the largest common elements. More...
 
template<class T >
rankAtomsByRank (const RDKit::ROMol &mol, const T &commAtms, bool ascending=true)
 Sort a list of atoms by their CIP rank. More...
 
double computeSubAngle (unsigned int degree, RDKit::Atom::HybridizationType htype)
 computes a subangle for an atom of given hybridization and degree More...
 
int rotationDir (const RDGeom::Point2D &center, const RDGeom::Point2D &loc1, const RDGeom::Point2D &loc2, double remAngle)
 computes the rotation direction between two vectors More...
 
RDGeom::Point2D computeNormal (const RDGeom::Point2D &center, const RDGeom::Point2D &other)
 computes and return the normal of a vector between two points More...
 
double computeAngle (const RDGeom::Point2D &center, const RDGeom::Point2D &loc1, const RDGeom::Point2D &loc2)
 computes the rotation angle between two vectors More...
 
int pickFirstRingToEmbed (const RDKit::ROMol &mol, const RDKit::VECT_INT_VECT &fusedRings)
 pick the ring to embed first in a fused system More...
 
RDKit::INT_VECT getRotatableBonds (const RDKit::ROMol &mol, unsigned int aid1, unsigned int aid2)
 find the rotatable bonds on the shortest path between two atoms we will ignore ring atoms, and double bonds which are marked cis/trans More...
 
RDKit::INT_VECT getAllRotatableBonds (const RDKit::ROMol &mol)
 find all the rotatable bonds in a molecule we will ignore ring atoms, and double bonds which are marked cis/trans More...
 
void getNbrAtomAndBondIds (unsigned int aid, const RDKit::ROMol *mol, RDKit::INT_VECT &aids, RDKit::INT_VECT &bids)
 Get the ids of the atoms and bonds that are connected to aid. More...
 
INT_PAIR_VECT findBondsPairsToPermuteDeg4 (const RDGeom::Point2D &center, const RDKit::INT_VECT &nbrBids, const VECT_C_POINT &nbrLocs)
 Find pairs of bonds that can be permuted at a non-ring degree 4 atom. More...
 
int getAtomDepictRank (const RDKit::Atom *at)
 returns the rank of the atom for determining draw order More...
 
unsigned int compute2DCoords (RDKit::ROMol &mol, const RDGeom::INT_POINT2D_MAP *coordMap=0, bool canonOrient=false, bool clearConfs=true, unsigned int nFlipsPerSample=0, unsigned int nSamples=0, int sampleSeed=0, bool permuteDeg4Nodes=false)
 Generate 2D coordinates (a depiction) for a molecule. More...
 
unsigned int compute2DCoordsMimicDistMat (RDKit::ROMol &mol, const DOUBLE_SMART_PTR *dmat=0, bool canonOrient=true, bool clearConfs=true, double weightDistMat=0.5, unsigned int nFlipsPerSample=3, unsigned int nSamples=100, int sampleSeed=25, bool permuteDeg4Nodes=true)
 Compute the 2D coordinates such the interatom distances. More...
 

Variables

double BOND_LEN
 
double COLLISION_THRES
 
double BOND_THRES
 
double ANGLE_OPEN
 
unsigned int MAX_COLL_ITERS
 
double HETEROATOM_COLL_SCALE
 
unsigned int NUM_BONDS_FLIPS
 

Typedef Documentation

typedef std::pair<double, INT_PAIR> RDDepict::DOUBLE_INT_PAIR

Definition at line 162 of file DepictUtils.h.

typedef boost::shared_array<double> RDDepict::DOUBLE_SMART_PTR

Definition at line 25 of file EmbeddedFrag.h.

typedef std::map<unsigned int, EmbeddedAtom> RDDepict::INT_EATOM_MAP

Definition at line 133 of file EmbeddedFrag.h.

typedef INT_EATOM_MAP::const_iterator RDDepict::INT_EATOM_MAP_CI

Definition at line 135 of file EmbeddedFrag.h.

typedef INT_EATOM_MAP::iterator RDDepict::INT_EATOM_MAP_I

Definition at line 134 of file EmbeddedFrag.h.

typedef std::pair<int, int> RDDepict::INT_PAIR

Definition at line 158 of file DepictUtils.h.

typedef std::vector<INT_PAIR> RDDepict::INT_PAIR_VECT

Definition at line 159 of file DepictUtils.h.

typedef INT_PAIR_VECT::const_iterator RDDepict::INT_PAIR_VECT_CI

Definition at line 160 of file DepictUtils.h.

typedef std::list<PAIR_D_I_I> RDDepict::LIST_PAIR_DII

Definition at line 44 of file DepictUtils.h.

typedef std::pair<double, PAIR_I_I> RDDepict::PAIR_D_I_I

Definition at line 43 of file DepictUtils.h.

typedef std::pair<int, int> RDDepict::PAIR_I_I

Definition at line 33 of file DepictUtils.h.

typedef std::priority_queue<PAIR_I_I, VECT_PII, gtIIPair> RDDepict::PR_QUEUE

Definition at line 41 of file DepictUtils.h.

typedef std::vector<const RDGeom::Point2D *> RDDepict::VECT_C_POINT

Definition at line 31 of file DepictUtils.h.

typedef std::vector<PAIR_I_I> RDDepict::VECT_PII

Definition at line 34 of file DepictUtils.h.

Function Documentation

unsigned int RDDepict::compute2DCoords ( RDKit::ROMol mol,
const RDGeom::INT_POINT2D_MAP coordMap = 0,
bool  canonOrient = false,
bool  clearConfs = true,
unsigned int  nFlipsPerSample = 0,
unsigned int  nSamples = 0,
int  sampleSeed = 0,
bool  permuteDeg4Nodes = false 
)

Generate 2D coordinates (a depiction) for a molecule.

Parameters
molthe molecule were are interested in
coordMapa map of int to Point2D, between atom IDs and their locations. This is the container the user needs to fill if he/she wants to specify coordinates for a portion of the molecule, defaults to 0
canonOrientcanonicalize the orientation so that the long axes align with the x-axis etc.
clearConfsclear all existing conformations on the molecule before adding the 2D coordinates instead of simply adding to the list
nFlipsPerSample- the number of rotatable bonds that are flipped at random for each sample
nSamples- the number of samples
sampleSeed- seed for the random sampling process
permuteDeg4Nodes- try permuting the drawing order of bonds around atoms with four neighbors in order to improve the depiction
Returns
ID of the conformation added to the molecule cotaining the 2D coordinates

Referenced by RDKit::Drawing::MolToDrawing().

void RDDepict::compute2DCoordsForReaction ( RDKit::ChemicalReaction rxn,
double  spacing = 2.0,
bool  updateProps = true,
bool  canonOrient = false,
unsigned int  nFlipsPerSample = 0,
unsigned int  nSamples = 0,
int  sampleSeed = 0,
bool  permuteDeg4Nodes = false 
)

Generate 2D coordinates (a depiction) for a reaction.

Parameters
rxnthe reaction we are interested in
spacingthe spacing between components of the reaction
updatePropsif set, properties such as conjugation and hybridization will be calculated for the reactant and product templates before generating coordinates. This should result in better depictions, but can lead to errors in some cases.
canonOrientcanonicalize the orientation so that the long axes align with the x-axis etc.
nFlipsPerSample- the number of rotatable bonds that are flipped at random for each sample
nSamples- the number of samples
sampleSeed- seed for the random sampling process
permuteDeg4Nodes- try permuting the drawing order of bonds around atoms with four neighbors in order to improve the depiction

for the other parameters see the documentation for compute2DCoords()

unsigned int RDDepict::compute2DCoordsMimicDistMat ( RDKit::ROMol mol,
const DOUBLE_SMART_PTR dmat = 0,
bool  canonOrient = true,
bool  clearConfs = true,
double  weightDistMat = 0.5,
unsigned int  nFlipsPerSample = 3,
unsigned int  nSamples = 100,
int  sampleSeed = 25,
bool  permuteDeg4Nodes = true 
)

Compute the 2D coordinates such the interatom distances.

This function generates 2D coordinates such that the inter-atom distances mimic those specified via dmat. This is done by randomly sampling(flipping) the rotatable bonds in the molecule and evaluating a cost function which contains two components. The first component is the sum of inverse of the squared inter-atom distances, this helps in spreading the atoms far from each other. The second component is the sum of squares of the difference in distance between those in dmat and the generated structure. The user can adjust the relative importance of the two components via a adjustable paramter (see below)

ARGUMENTS:

Parameters
mol- molecule to generate coordinates for
dmat- the distance matrix we want to mimic, this is a symmetric N by N matrix where N is the number of atoms in mol. All negative entries in dmat are ignored.
canonOrient- canonicalize the orientation after the 2D embedding is done
clearConfs- clear any previously existing conformations on mol before adding a conformation
weightDistMat- A value between 0.0 and 1.0, this determines the importance of mimicing the the inter atoms distances in dmat. (1.0 - weightDistMat) is the weight associated to spreading out the structure (density) in the cost function
nFlipsPerSample- the number of rotatable bonds that are flipped at random for each sample
nSamples- the number of samples
sampleSeed- seed for the random sampling process
permuteDeg4Nodes- try permuting the drawing order of bonds around atoms with four neighbors in order to improve the depiction
Returns
ID of the conformation added to the molecule cotaining the 2D coordinates
double RDDepict::computeAngle ( const RDGeom::Point2D center,
const RDGeom::Point2D loc1,
const RDGeom::Point2D loc2 
)
inline

computes the rotation angle between two vectors

Parameters
centerthe common point
loc1endpoint 1
loc2endpoint 2
Returns
the angle (in radians)

Definition at line 270 of file DepictUtils.h.

References RDGeom::Point2D::angleTo(), findBondsPairsToPermuteDeg4(), getAllRotatableBonds(), getNbrAtomAndBondIds(), getRotatableBonds(), and pickFirstRingToEmbed().

RDGeom::Point2D RDDepict::computeBisectPoint ( const RDGeom::Point2D rcr,
double  ang,
const RDGeom::Point2D nb1,
const RDGeom::Point2D nb2 
)

Find a point that bisects the angle at rcr.

The new point lies between nb1 and nb2. The line (rcr, newPt) bisects the angle 'ang' at rcr

RDGeom::Point2D RDDepict::computeNormal ( const RDGeom::Point2D center,
const RDGeom::Point2D other 
)
inline

computes and return the normal of a vector between two points

Parameters
centerthe common point
otherthe endpoint
Returns
the normal

Definition at line 252 of file DepictUtils.h.

References RDGeom::Point2D::normalize(), RDGeom::Point2D::x, and RDGeom::Point2D::y.

double RDDepict::computeSubAngle ( unsigned int  degree,
RDKit::Atom::HybridizationType  htype 
)
inline

computes a subangle for an atom of given hybridization and degree

Parameters
degreethe degree of the atom (number of neighbors)
htypethe atom's hybridization
Returns
the subangle (in radians)

Definition at line 181 of file DepictUtils.h.

References M_PI, RDKit::Atom::SP2, RDKit::Atom::SP3, and RDKit::Atom::UNSPECIFIED.

RDGeom::INT_POINT2D_MAP RDDepict::embedRing ( const RDKit::INT_VECT ring)

Some utility functions used in generating 2D coordinates.

Embed a ring as a convex polygon in 2D

The process here is very straightforward:

We take the center of the ring to lie at the origin, so put the first point at the origin and then sweep anti-clockwise by an angle A = 360/n for the next point.

The length of the arm (l) we want to sweep is easy to compute given the bond length (b) we want to use for each bond in the ring (for now we will assume that this bond legnth is the same for all bonds in the ring:

l = b/sqrt(2*(1 - cos(A))

the above formula derives from the triangle formula, where side 'c' is given in terms of sides 'a' and 'b' as:

c = a^2 + b^2 - 2.a.b.cos(A)

where A is the angle between a and b

INT_PAIR_VECT RDDepict::findBondsPairsToPermuteDeg4 ( const RDGeom::Point2D center,
const RDKit::INT_VECT nbrBids,
const VECT_C_POINT nbrLocs 
)

Find pairs of bonds that can be permuted at a non-ring degree 4 atom.

This function will return only those pairs that cannot be permuted by flipping a rotatble bond

   D
   |
   b3
   |

A-b1-B-b2-C | b4 | E For example in teh above situation on the pairs (b1, b3) and (b1, b4) will be returned All other permutations can be achieved via a rotatable bond flip.

ARGUMENTS:

Parameters
center- location of the central atom
nbrBids- a vector (of length 4) containing the ids of the bonds to the neighbors
nbrLocs- locations of the neighbors

Referenced by computeAngle().

RDKit::INT_VECT RDDepict::findNextRingToEmbed ( const RDKit::INT_VECT doneRings,
const RDKit::VECT_INT_VECT fusedRings,
int &  nextId 
)

From a given set of rings find the ring the largest common elements.

RDKit::INT_VECT RDDepict::getAllRotatableBonds ( const RDKit::ROMol mol)

find all the rotatable bonds in a molecule we will ignore ring atoms, and double bonds which are marked cis/trans

Note that rotatable in this context doesn't connect to the standard chemical definition of a rotatable bond; we're just talking about bonds than can be flipped in order to clean up the depiction.

Parameters
molthe molecule of interest
Returns
a set of the indices of the rotatable bonds

Referenced by computeAngle().

int RDDepict::getAtomDepictRank ( const RDKit::Atom at)
inline

returns the rank of the atom for determining draw order

Definition at line 350 of file DepictUtils.h.

References RDKit::Atom::getAtomicNum(), and RDKit::Atom::getDegree().

void RDDepict::getNbrAtomAndBondIds ( unsigned int  aid,
const RDKit::ROMol mol,
RDKit::INT_VECT aids,
RDKit::INT_VECT bids 
)

Get the ids of the atoms and bonds that are connected to aid.

Referenced by computeAngle().

RDKit::INT_VECT RDDepict::getRotatableBonds ( const RDKit::ROMol mol,
unsigned int  aid1,
unsigned int  aid2 
)

find the rotatable bonds on the shortest path between two atoms we will ignore ring atoms, and double bonds which are marked cis/trans

Note that rotatable in this context doesn't connect to the standard chemical definition of a rotatable bond; we're just talking about bonds than can be flipped in order to clean up the depiction.

Parameters
molthe molecule of interest
aid1index of the first atom
aid2index of the second atom
Returns
a set of the indices of the rotatable bonds

Referenced by computeAngle().

int RDDepict::pickFirstRingToEmbed ( const RDKit::ROMol mol,
const RDKit::VECT_INT_VECT fusedRings 
)

pick the ring to embed first in a fused system

Parameters
molthe molecule of interest
fusedRingsthe collection of the molecule's fused rings
Returns
the index of the ring with the least number of substitutions

Referenced by computeAngle().

template<class T >
T RDDepict::rankAtomsByRank ( const RDKit::ROMol mol,
const T &  commAtms,
bool  ascending = true 
)

Sort a list of atoms by their CIP rank.

Parameters
molmolecule of interest
commAtmsatoms that need to be ranked
ascendingsort to an ascending order or a descending order
RDGeom::Point2D RDDepict::reflectPoint ( const RDGeom::Point2D point,
const RDGeom::Point2D loc1,
const RDGeom::Point2D loc2 
)
void RDDepict::reflectPoints ( RDGeom::INT_POINT2D_MAP coordMap,
const RDGeom::Point2D loc1,
const RDGeom::Point2D loc2 
)

Reflect a set of point through a the line joining two point.

ARGUMENTS:

Parameters
coordMapa map of <int, point2D> going from atom id to current coordinates of the points that need to be reflected: The coordinates are overwritten
loc1the first point of the line that is to be used as a mirror
loc2the second point of the line to be used as a mirror
int RDDepict::rotationDir ( const RDGeom::Point2D center,
const RDGeom::Point2D loc1,
const RDGeom::Point2D loc2,
double  remAngle 
)
inline

computes the rotation direction between two vectors

Let:

v1 = loc1 - center

v2 = loc2 - center

If remaining angle(v1, v2) is < 180 and corss(v1, v2) > 0.0 then the rotation dir is +1.0

else if remAngle(v1, v2) is > 180 and cross(v1, v2) < 0.0 then rotation dir is -1.0

else if remAngle(v1, v2) is < 180 and cross(v1, v2) < 0.0 then rotation dir is -1.0

finally if remAngle(v1, v2) is > 180 and cross(v1, v2) < 0.0 then rotation dir is +1.0

Parameters
centerthe common point
loc1endpoint 1
loc2endpoint 2
remAnglethe remaining angle about center in radians
Returns
the rotation direction (1 or -1)

Definition at line 230 of file DepictUtils.h.

References M_PI, RDGeom::Point2D::x, and RDGeom::Point2D::y.

RDKit::INT_VECT RDDepict::setNbrOrder ( unsigned int  aid,
const RDKit::INT_VECT nbrs,
const RDKit::ROMol mol 
)

Set the neighbors yet to added to aid such that the atoms with the most subs.

Ok this needs some explanation

  • Let A, B, C, D be the substituent on the central atom X (given by atom index aid)
  • also let A be the atom that is already embedded
  • Now we want the order in which the remaining neighbors B,C,D are added to X such that the atoms with atom with largest number of substituent fall on opposite sides of X so as to minimize atom clashes later in the depiction

E.g. let say we have the following situation

          B
       |  |
       A--X--C
       |  |
        --D--
          |

In this case the the number substituent of A, B, C, D are 3, 1, 1, 4 respectively so want to A and D to go opposite sides and so that we draw

          B
       |  |  |
       A--X--D--
       |  |  |
          C

And the correct ordering of the neighbors is B,D,C

void RDDepict::transformPoints ( RDGeom::INT_POINT2D_MAP nringCor,
const RDGeom::Transform2D trans 
)

Variable Documentation

double RDDepict::ANGLE_OPEN
double RDDepict::BOND_LEN
double RDDepict::BOND_THRES
double RDDepict::COLLISION_THRES
double RDDepict::HETEROATOM_COLL_SCALE
unsigned int RDDepict::MAX_COLL_ITERS
unsigned int RDDepict::NUM_BONDS_FLIPS