Re: [Rdkit-discuss] how to exclude conformers with wrong geometries
Hi Greg, Actually this question is relevant to my recent thread about aligning MCS of 2 compounds. As also seen in the code I have posted in my last email, I first generate N conformers of each query compound and then optimize them using distance restraints in order to superimpose their MCS with that of the reference ligand. So I tried the new code as you suggested and although less, I still get wrong geometries for some compounds, particularly in the aromatic rings. I tried to repair the wrong conformers by copying the original bondlengths, angle and dihedral values from my starting conformation which always has correct geometry. However I get the following errors whenever I encounter aromatic atoms: from Chem.rdMolTransforms.SetDihedralDeg() > ValueError: bond (j,k) must not belong to a ring > from Chem.rdMolTransforms.SetAngleDeg() > ValueError: bonds (i,j) and (j,k) must not both belong to a ring > from Chem.rdMolTransforms.SetBondLength() > ValueError: bond (i,j) must not belong to a ring > How can I reset the bondlengths, angles and dihedrals of the aromatic rings to their initial values? thanks Thomas On 23 February 2017 at 20:52, Greg Landrum wrote: > One thing you both might find useful is generating conformers using the > new ETKDG code. The parameters for this are described in the documentation. > Look for the "Working with 3D molecules" section of the getting started > guide. > > That code should solve most, if not all, of the problems you are > discussing. > > -greg > > > -- Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Re: [Rdkit-discuss] how to exclude conformers with wrong geometries
One thing you both might find useful is generating conformers using the new ETKDG code. The parameters for this are described in the documentation. Look for the "Working with 3D molecules" section of the getting started guide. That code should solve most, if not all, of the problems you are discussing. -greg On Thu, 23 Feb 2017 at 15:44, Michal Krompiec wrote: > Hi Thomas, > I had a similar problem, but in my case checking for unusually long bonds > (e.g. longer than typical bond length times an experimental multiplier) and > "additional" close contacts, i.e. more close contacts (excluding h-bonds) > between atoms separated by at least 4 bonds (if I remember correctly) than > in a conformer without restraints did the trick. You may also want to tweak > the force constant of the restraints. > Kind regards, > Michal > > On Thu, 23 Feb 2017 at 13:50, Thomas Evangelidis > wrote: > > Greetings, > > I want to create conformations of molecules using restraints but many of > the resulting conformers have wrong geometries (distorted aromatic, rings, > out-of-plane hydrogens, etc.). I would like to define some rules of thump > to recognize such wrong geometries. One possible way is to look at the > total energy of the molecules as given by > AllChem.UFFGetMoleculeForceField.CalcEnergy(). However, the energy > threshold is different for every molecule. Therefore I thought to define an > arbitrary threshold dependent on the total number of heavy atoms, like N*20 > kcal/mol. When all the generated conformers exceed that energy thresholds, > then more conformers are generated until I get at least one with valid > energy. I would like to know your opinion on this matter. What would be a > reasonable multiplier (in my example I used 20 kcal/mol)? Would you > recommend adding other factors (e.g. number of rotatable bonds) in my > empirical rule to exclude wrong geometries? > > Another way could be to apply bond and dihedral angle constrains to the > aromatic atoms based on their configuration in the starting conformation > (which has always correct geometry). But that would be more tedious than > the energy threshold approach. Do you think that this is the right way to > go? > > Thanks in advance. > Thomas > > > -- > > == > > Thomas Evangelidis > > Research Specialist > CEITEC - Central European Institute of Technology > Masaryk University > Kamenice 5/A35/1S081, > 62500 Brno, Czech Republic > > email: tev...@pharm.uoa.gr > > teva...@gmail.com > > > website: https://sites.google.com/site/thomasevangelidishomepage/ > > > -- > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > ___ > Rdkit-discuss mailing list > Rdkit-discuss@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/rdkit-discuss > > > -- > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > ___ > Rdkit-discuss mailing list > Rdkit-discuss@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/rdkit-discuss > -- Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Re: [Rdkit-discuss] how to exclude conformers with wrong geometries
Hi Thomas, I had a similar problem, but in my case checking for unusually long bonds (e.g. longer than typical bond length times an experimental multiplier) and "additional" close contacts, i.e. more close contacts (excluding h-bonds) between atoms separated by at least 4 bonds (if I remember correctly) than in a conformer without restraints did the trick. You may also want to tweak the force constant of the restraints. Kind regards, Michal On Thu, 23 Feb 2017 at 13:50, Thomas Evangelidis wrote: > Greetings, > > I want to create conformations of molecules using restraints but many of > the resulting conformers have wrong geometries (distorted aromatic, rings, > out-of-plane hydrogens, etc.). I would like to define some rules of thump > to recognize such wrong geometries. One possible way is to look at the > total energy of the molecules as given by > AllChem.UFFGetMoleculeForceField.CalcEnergy(). However, the energy > threshold is different for every molecule. Therefore I thought to define an > arbitrary threshold dependent on the total number of heavy atoms, like N*20 > kcal/mol. When all the generated conformers exceed that energy thresholds, > then more conformers are generated until I get at least one with valid > energy. I would like to know your opinion on this matter. What would be a > reasonable multiplier (in my example I used 20 kcal/mol)? Would you > recommend adding other factors (e.g. number of rotatable bonds) in my > empirical rule to exclude wrong geometries? > > Another way could be to apply bond and dihedral angle constrains to the > aromatic atoms based on their configuration in the starting conformation > (which has always correct geometry). But that would be more tedious than > the energy threshold approach. Do you think that this is the right way to > go? > > Thanks in advance. > Thomas > > > -- > > == > > Thomas Evangelidis > > Research Specialist > CEITEC - Central European Institute of Technology > Masaryk University > Kamenice 5/A35/1S081, > 62500 Brno, Czech Republic > > email: tev...@pharm.uoa.gr > > teva...@gmail.com > > > website: https://sites.google.com/site/thomasevangelidishomepage/ > > > -- > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > ___ > Rdkit-discuss mailing list > Rdkit-discuss@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/rdkit-discuss > -- Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot___ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
