Re: [Rdkit-discuss] 3D alignment in Python: align conformers of 2 molecules
Hi all,
here the code for Open3DALIGN:
cids = generateConformers(mol, numConformers)
prbPyMP = AllChem.MMFFGetMoleculeProperties(mol)
maxScore = 0;
for refCid in refCids:
for cid in cids:
alignment = AllChem.GetO3A(mol, refMol, prbPyMP, refPyMP,
prbCid=cid, refCid=refCid,)
score = alignment.Score()
logger.debug('Score: %.2f', score)
if score maxScore:
logger.info('New max. Score: %.2f', score)
maxScore = score
refConformerId = refCid
molCid = cid
alignment = AllChem.GetO3A(mol, refMol, prbPyMP, refPyMP,
prbCid=molCid, refCid=refConformerId)
alignment.Align()
# show in PyMol
One question I have is how I can align this to a specific fragment or the MCS?
I tried using constraintMap but for some reason o3a does not want to align the
2 structure on it. I suspect because then the rest of the alignment will not be
very good in that case.
The code:
constraintMap = []
constraintWeights = []
mcs = MCS.FindMCS([refMol,mol], bondCompare='bondtypes',
ringMatchesRingOnly=False)
if mcs.completed == 1 and mcs.numAtoms 0:
core = Chem.MolFromSmarts(mcs.smarts) # or use specific smarts
pattern through argument option
logger.info('MCS: %s', Chem.MolToSmiles(core))
refMatch = refMol.GetSubstructMatch(core)
match = mol.GetSubstructMatch(core)
for idx, val in enumerate(match):
constraintMap.append((val, refMatch[idx]))
constraintWeights.append(100.0)
logger.info(constraintMap)
...
alignment = AllChem.GetO3A(mol, refMol, prbPyMP, refPyMP, prbCid=cid,
refCid=refCid, constraintMap=constraintMap,
constraintWeights=constraintWeights)
The alinment is different but it still does not want to align to my defined
fragment. even if I set constraintWeights to very high value. In my case I
especially want to align heteroatoms properly. For that I tried:
for idx, val in enumerate(match):
constraintMap.append((val, refMatch[idx]))
if mol.GetAtomWithIdx(val).GetAtomicNum() != 6:
constraintWeights.append(1000.0)
logger.info('Set weigth to 1000')
else:
constraintWeights.append(10.0)
But the single hetero atoms in this my test case are still not aligned on each
other. I also tried ConstrainedEmbed (or how it is called) but that results in
either an error or with relaxed parameters in a completely useless
conformation, namely on a ring.
What are my options?
From: greg.land...@gmail.com
Date: Fri, 27 Jun 2014 08:20:56 +0200
Subject: Re: [Rdkit-discuss] 3D alignment in Python: align conformers of 2
molecules
To: beginn...@hotmail.de
CC: rdkit-discuss@lists.sourceforge.net
On Fri, Jun 27, 2014 at 7:57 AM, Thomas Strunz beginn...@hotmail.de wrote:
thanks for your quick reply. This helped to improve the alignment.
I'm glad to hear it!
How can I reproduce the alignment done in with the Open3DAlign Node in Python?
Is it possible at all?
But of course. :-)There's some example code that shows how to do it on page 37
of Paolo's presentation from the last RDKit UGM:
https://github.com/rdkit/UGM_2013/raw/master/Presentations/Tosco.RDKit_UGM2013.pdf
-greg
--
Open source business process management suite built on Java and Eclipse
Turn processes into business applications with Bonita BPM Community Edition
Quickly connect people, data, and systems into organized workflows
Winner of BOSSIE, CODIE, OW2 and Gartner awards
http://p.sf.net/sfu/Bonitasoft___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Re: [Rdkit-discuss] 3D alignment in Python: align conformers of 2 molecules
Dear Thomas,
if you wish to align two structures by their MCS, O3A is probably not
the tool for you. O3A is meant for unsupervised alignment, and it will
attempt to align two structures matching the most similar pairs of atoms
between the two. Similarity is defined by the closeness of their MMFF94
atom types and charges. Constraints force the addition of certain atom
pairs in the atom map which is built in an unsupervised fashion. Whether
or not such constraints succeed in giving you the desired alignment
depends on their weight compared to the rest of the atom map.
I think you have two options:
1) Use the AlignMol()method after having identified the MCSs/fragments
of interest in the 2 molecules and using them as atomMap
2) Try to carry out the alignment with O3A in two steps:
a) constrained alignment, global search (options = 0): this will give
you an initial
alignment,which may be a bit skewed by the high constraint weight and
thus sub-optimal
b) unconstrained alignment, local-only search(options = 4): this
might succeed in snapping
in place correctly matching atoms, provided that the initial
alignment is
good enough to allow the local optimization to do its job
HTH,
Paolo
On 01/07/14 09:37, Thomas Strunz wrote:
Hi all,
here the code for Open3DALIGN:
cids = generateConformers(mol, numConformers)
prbPyMP = AllChem.MMFFGetMoleculeProperties(mol)
maxScore = 0;
for refCid in refCids:
for cid in cids:
alignment = AllChem.GetO3A(mol, refMol, prbPyMP, refPyMP,
prbCid=cid, refCid=refCid,)
score = alignment.Score()
logger.debug('Score: %.2f', score)
if score maxScore:
logger.info('New max. Score: %.2f', score)
maxScore = score
refConformerId = refCid
molCid = cid
alignment = AllChem.GetO3A(mol, refMol, prbPyMP, refPyMP,
prbCid=molCid, refCid=refConformerId)
alignment.Align()
# show in PyMol
One question I have is how I can align this to a specific fragment or
the MCS? I tried using constraintMap but for some reason o3a does not
want to align the 2 structure on it. I suspect because then the rest
of the alignment will not be very good in that case.
The code:
constraintMap = []
constraintWeights = []
mcs = MCS.FindMCS([refMol,mol], bondCompare='bondtypes',
ringMatchesRingOnly=False)
if mcs.completed == 1 and mcs.numAtoms 0:
core = Chem.MolFromSmarts(mcs.smarts) # or use specific
smarts pattern through argument option
logger.info('MCS: %s', Chem.MolToSmiles(core))
refMatch = refMol.GetSubstructMatch(core)
match = mol.GetSubstructMatch(core)
for idx, val in enumerate(match):
constraintMap.append((val, refMatch[idx]))
constraintWeights.append(100.0)
logger.info(constraintMap)
...
alignment = AllChem.GetO3A(mol, refMol, prbPyMP, refPyMP,
prbCid=cid, refCid=refCid, constraintMap=constraintMap,
constraintWeights=constraintWeights)
The alinment is different but it still does not want to align to my
defined fragment. even if I set constraintWeights to very high value.
In my case I especially want to align heteroatoms properly. For that I
tried:
for idx, val in enumerate(match):
constraintMap.append((val, refMatch[idx]))
if mol.GetAtomWithIdx(val).GetAtomicNum() != 6:
constraintWeights.append(1000.0)
logger.info('Set weigth to 1000')
else:
constraintWeights.append(10.0)
But the single hetero atoms in this my test case are still not aligned
on each other. I also tried ConstrainedEmbed (or how it is called)
but that results in either an error or with relaxed parameters in a
completely useless conformation, namely on a ring.
What are my options?
From: greg.land...@gmail.com
Date: Fri, 27 Jun 2014 08:20:56 +0200
Subject: Re: [Rdkit-discuss] 3D alignment in Python: align conformers
of 2 molecules
To: beginn...@hotmail.de
CC: rdkit-discuss@lists.sourceforge.net
On Fri, Jun 27, 2014 at 7:57 AM, Thomas Strunz beginn...@hotmail.de
mailto:beginn...@hotmail.de wrote:
thanks for your quick reply. This helped to improve the alignment.
I'm glad to hear it!
How can I reproduce the alignment done in with the Open3DAlign
Node in Python? Is it possible at all?
But of course. :-)
There's some example code that shows how to do it on page 37 of
Paolo's presentation from the last RDKit UGM:
https://github.com/rdkit/UGM_2013/raw/master/Presentations/Tosco.RDKit_UGM2013.pdf
-greg
--
Open source business process
Re: [Rdkit-discuss] 3D alignment in Python: align conformers of 2 molecules
Hi Greg,
thanks for your quick reply. This helped to improve the alignment.
How can I reproduce the alignment done in with the Open3DAlign Node in Python?
Is it possible at all?
Best Regards,
Thomas
From: greg.land...@gmail.com
Date: Fri, 27 Jun 2014 05:10:42 +0200
Subject: Re: [Rdkit-discuss] 3D alignment in Python: align conformers of 2
molecules
To: beginn...@hotmail.de
CC: rdkit-discuss@lists.sourceforge.net
Hi Thomas,
I think there are a couple of problems with the code here:1) you aren't storing
the confirmation of the conformer of mol that produces the best alignment
(cid in the above code)
2) you aren't keeping the best alignment since you repeatedly run the same
molecule instances through the alignment code. You could fix this by, after the
double conformer loop, adding something like:
AllChem.AlignMol(refMol, mol, prbCid=refConformerId, refCid=molConformerId,
atomMap=zip(refMatch,match))
where I have assumed that molConformerId is the variable you use to solve
problem 1.
The loop would become something like this:
minRmsd = 1000; for refCid in refCids:
for cid in cids:
rmsd = AllChem.AlignMol(refMol, mol, prbCid=refCid, refCid=cid,
atomMap=zip(refMatch,match))if rmsd minRmsd:
minRmsd = rmsdrefConformerId = refCid
molConformerId = cid
rmsd = AllChem.AlignMol(refMol, mol, prbCid=refConformerId,
refCid=molConformerId,
atomMap=zip(refMatch,match))
It's also not really right to compare the results of this, which uses the MCS
code to find a common set of atoms and then does an RMSD alignment of those,
and the Open3DAlign results, which use a fuzzier scheme to identify the
atom-atom mapping between the molecules.
-greg
On Thu, Jun 26, 2014 at 2:32 PM, Thomas Strunz beginn...@hotmail.de wrote:
I'm trying to align all conformers of 2 molecules (and keep the best ones)
using the python api by following some of the tutorials:
http://nbviewer.ipython.org/gist/greglandrum/4316435/Working%20in%203D.ipynb
http://nbviewer.ipython.org/github/greglandrum/rdkit_blog/blob/master/notebooks/Using%20ConstrainedEmbed.ipynb
However whatever I try the alignment is considerably different than the one
created using the Open 3D alignment node in Knime.
Currently the issue is that the molecules do not seem to be aligned properly.
There is some sort of small shift. See code below.
Best Regards,
Thomas
refCids = generateConformers(refMol, numConformers)
mcs = MCS.FindMCS([refMol,mol], ringMatchesRingOnly=matchesRingOnly)
if mcs.completed == 1 and mcs.numAtoms 0:
core = Chem.MolFromSmarts(mcs.smarts)
logger.info('MCS: %s', Chem.MolToSmiles(core))
refMatch = refMol.GetSubstructMatch(core)
match = mol.GetSubstructMatch(core)
# conformers for current target
cids = generateConformers(mol, numConformers, coordMap=coordMap)
minRmsd = 1000;
for refCid in refCids:
for cid in cids:
rmsd = AllChem.AlignMol(refMol, mol, prbCid=refCid,
refCid=cid, atomMap=zip(refMatch,match))
logger.debug('RMSD: %.2f', rmsd)
if rmsd minRmsd:
logger.debug('New min RMSD: %.2f', rmsd)
minRmsd = rmsd
refConformerId = refCid
def generateConformers(mol, numConformers):
AllChem.EmbedMolecule(mol)
AllChem.MMFFOptimizeMolecule(mol)
cids=AllChem.EmbedMultipleConfs(mol, numConfs=numConformers,
maxAttempts=50, pruneRmsThresh=0.5, coordMap=coordMap)
for cid in cids: AllChem.MMFFOptimizeMolecule(mol,confId=cid)
return cids
--
Open source business process management suite built on Java and Eclipse
Turn processes into business applications with Bonita BPM Community Edition
Quickly connect people, data, and systems into organized workflows
Winner of BOSSIE, CODIE, OW2 and Gartner awards
http://p.sf.net/sfu/Bonitasoft
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
--
Open source business process management suite built on Java and Eclipse
Turn processes into business applications with Bonita BPM Community Edition
Quickly connect people, data, and systems into organized workflows
Winner of BOSSIE, CODIE, OW2 and Gartner awards
http://p.sf.net/sfu
Re: [Rdkit-discuss] 3D alignment in Python: align conformers of 2 molecules
On Fri, Jun 27, 2014 at 7:57 AM, Thomas Strunz beginn...@hotmail.de wrote: thanks for your quick reply. This helped to improve the alignment. I'm glad to hear it! How can I reproduce the alignment done in with the Open3DAlign Node in Python? Is it possible at all? But of course. :-) There's some example code that shows how to do it on page 37 of Paolo's presentation from the last RDKit UGM: https://github.com/rdkit/UGM_2013/raw/master/Presentations/Tosco.RDKit_UGM2013.pdf -greg -- Open source business process management suite built on Java and Eclipse Turn processes into business applications with Bonita BPM Community Edition Quickly connect people, data, and systems into organized workflows Winner of BOSSIE, CODIE, OW2 and Gartner awards http://p.sf.net/sfu/Bonitasoft___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
[Rdkit-discuss] 3D alignment in Python: align conformers of 2 molecules
I'm trying to align all conformers of 2 molecules (and keep the best ones)
using the python api by following some of the tutorials:
http://nbviewer.ipython.org/gist/greglandrum/4316435/Working%20in%203D.ipynb
http://nbviewer.ipython.org/github/greglandrum/rdkit_blog/blob/master/notebooks/Using%20ConstrainedEmbed.ipynb
However whatever I try the alignment is considerably different than the one
created using the Open 3D alignment node in Knime.
Currently the issue is that the molecules do not seem to be aligned properly.
There is some sort of small shift. See code below.
Best Regards,
Thomas
refCids = generateConformers(refMol, numConformers)
mcs = MCS.FindMCS([refMol,mol], ringMatchesRingOnly=matchesRingOnly)
if mcs.completed == 1 and mcs.numAtoms 0:
core = Chem.MolFromSmarts(mcs.smarts)
logger.info('MCS: %s', Chem.MolToSmiles(core))
refMatch = refMol.GetSubstructMatch(core)
match = mol.GetSubstructMatch(core)
# conformers for current target
cids = generateConformers(mol, numConformers, coordMap=coordMap)
minRmsd = 1000;
for refCid in refCids:
for cid in cids:
rmsd = AllChem.AlignMol(refMol, mol, prbCid=refCid,
refCid=cid, atomMap=zip(refMatch,match))
logger.debug('RMSD: %.2f', rmsd)
if rmsd minRmsd:
logger.debug('New min RMSD: %.2f', rmsd)
minRmsd = rmsd
refConformerId = refCid
def generateConformers(mol, numConformers):
AllChem.EmbedMolecule(mol)
AllChem.MMFFOptimizeMolecule(mol)
cids=AllChem.EmbedMultipleConfs(mol, numConfs=numConformers,
maxAttempts=50, pruneRmsThresh=0.5, coordMap=coordMap)
for cid in cids: AllChem.MMFFOptimizeMolecule(mol,confId=cid)
return cids
--
Open source business process management suite built on Java and Eclipse
Turn processes into business applications with Bonita BPM Community Edition
Quickly connect people, data, and systems into organized workflows
Winner of BOSSIE, CODIE, OW2 and Gartner awards
http://p.sf.net/sfu/Bonitasoft___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Re: [Rdkit-discuss] 3D alignment in Python: align conformers of 2 molecules
Hi Thomas,
I think there are a couple of problems with the code here:
1) you aren't storing the confirmation of the conformer of mol that
produces the best alignment (cid in the above code)
2) you aren't keeping the best alignment since you repeatedly run the same
molecule instances through the alignment code. You could fix this by, after
the double conformer loop, adding something like:
AllChem.AlignMol(refMol, mol, prbCid=refConformerId,
refCid=molConformerId, atomMap=zip(refMatch,match))
where I have assumed that molConformerId is the variable you use to solve
problem 1.
The loop would become something like this:
minRmsd = 1000;
for refCid in refCids:
for cid in cids:
rmsd = AllChem.AlignMol(refMol, mol, prbCid=refCid,
refCid=cid, atomMap=zip(refMatch,match))
if rmsd minRmsd:
minRmsd = rmsd
refConformerId = refCid
molConformerId = cid
rmsd = AllChem.AlignMol(refMol, mol, prbCid=refConformerId,
refCid=molConformerId,
atomMap=zip(refMatch,match))
It's also not really right to compare the results of this, which uses the
MCS code to find a common set of atoms and then does an RMSD alignment of
those, and the Open3DAlign results, which use a fuzzier scheme to identify
the atom-atom mapping between the molecules.
-greg
On Thu, Jun 26, 2014 at 2:32 PM, Thomas Strunz beginn...@hotmail.de wrote:
I'm trying to align all conformers of 2 molecules (and keep the best ones)
using the python api by following some of the tutorials:
http://nbviewer.ipython.org/gist/greglandrum/4316435/Working%20in%203D.ipynb
http://nbviewer.ipython.org/github/greglandrum/rdkit_blog/blob/master/notebooks/Using%20ConstrainedEmbed.ipynb
However whatever I try the alignment is considerably different than the
one created using the Open 3D alignment node in Knime.
Currently the issue is that the molecules do not seem to be aligned
properly. There is some sort of small shift. See code below.
Best Regards,
Thomas
refCids = generateConformers(refMol, numConformers)
mcs = MCS.FindMCS([refMol,mol],
ringMatchesRingOnly=matchesRingOnly)
if mcs.completed == 1 and mcs.numAtoms 0:
core = Chem.MolFromSmarts(mcs.smarts)
logger.info('MCS: %s', Chem.MolToSmiles(core))
refMatch = refMol.GetSubstructMatch(core)
match = mol.GetSubstructMatch(core)
# conformers for current target
cids = generateConformers(mol, numConformers,
coordMap=coordMap)
minRmsd = 1000;
for refCid in refCids:
for cid in cids:
rmsd = AllChem.AlignMol(refMol, mol, prbCid=refCid,
refCid=cid, atomMap=zip(refMatch,match))
logger.debug('RMSD: %.2f', rmsd)
if rmsd minRmsd:
logger.debug('New min RMSD: %.2f', rmsd)
minRmsd = rmsd
refConformerId = refCid
def generateConformers(mol, numConformers):
AllChem.EmbedMolecule(mol)
AllChem.MMFFOptimizeMolecule(mol)
cids=AllChem.EmbedMultipleConfs(mol, numConfs=numConformers,
maxAttempts=50, pruneRmsThresh=0.5, coordMap=coordMap)
for cid in cids: AllChem.MMFFOptimizeMolecule(mol,confId=cid)
return cids
--
Open source business process management suite built on Java and Eclipse
Turn processes into business applications with Bonita BPM Community Edition
Quickly connect people, data, and systems into organized workflows
Winner of BOSSIE, CODIE, OW2 and Gartner awards
http://p.sf.net/sfu/Bonitasoft
___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
--
Open source business process management suite built on Java and Eclipse
Turn processes into business applications with Bonita BPM Community Edition
Quickly connect people, data, and systems into organized workflows
Winner of BOSSIE, CODIE, OW2 and Gartner awards
http://p.sf.net/sfu/Bonitasoft___
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
