When I was at Schrödinger, I wrote a simple program to find bad 2D structures. I no longer have access to the code, but I computed two things:
1. Number of bond lengths deviating from the median bond length (MBL) by 50% or more (i.e., <0.5*MBL or >2*MBL) 2. Number of bond crossings The overall score for a structure was the sum of these two numbers. This won't pick the best-looking structures when the above numbers are 0, but the the worst structures on a sorted list (one might almost say a sordid list) of 10K drug-like structures rendered by our then (pre-coordgen) 2D depiction sure were bad. -P. On Tue, Apr 9, 2019 at 10:44 AM Thomas Evangelidis <teva...@gmail.com> wrote: > Hello Lukas, > > I am also struggling with 2D coordinate generation quite a long time as > well as what criteria to use for choosing the most appropriate. Therefore, > I would be very interest to use your code for 2D coordinate selection. > > With best regards, > Thomas > > PS: very nice notebook Jose. I also wanted to write something similar but > never really found enough time to finish it. > > > > On Tue, 9 Apr 2019 at 16:31, Lukas Pravda <lpra...@ebi.ac.uk> wrote: > >> Hi Jose, >> >> As you have shown there is no single method which would be perfect for >> everything. If you don’t care that much about speed, the possible solution >> could be to compute coordinates with all three approaches and then simply >> select the best conformer based on some criteria. >> >> The solution I use is to generate 2D coordinates using multiple >> approaches and then I have a set of methods which computes number of bond >> collisions and atoms being close to each other using KD-tree. Altogether >> this all is expressed as penalty score, where the lower is better. >> >> Should you need any code, let me know. >> >> Lukas >> >> On 09/04/2019, 14:35, "Jose Manuel Gally" <jose.manuel.ga...@gmail.com> >> wrote: >> >> Dear all, >> >> This might sound naive, but I want to compute 2D coordinates for a >> set >> of molecules. >> >> For now I am considering the 3 methods below [1]. >> >> I was wondering if there was any recommendation to use one method >> over >> another in some cases? >> >> For instance, very large rings are not displayed round for CoordGen >> but >> sometimes this method performs worse than the default (AllChem). >> >> Computational time is not really an issue here as I generate those >> coordinates on the fly for a very small set of compounds. >> >> Here is a gist with a few examples: >> https://gist.github.com/jose-manuel/0f2a5e8eae8bf2a72c0faad7f2f2a263 >> >> Thanks in advance, any suggestion is welcome! >> >> Cheers, >> Jose Manuel >> >> [1] Methods: >> >> 1) rdkit.AllChem.Compute2dCoors (equivqlent to >> rdkit.Chem.rdDepictor.Compute2DCoords) >> 2) rdkit.Avalon.pyAvalonTools.Generate2DCoords >> 3) rdkit.Chem.rdCoordGen.AddCoords + rescale >> >> >> >> >> >> _______________________________________________ >> Rdkit-discuss mailing list >> Rdkit-discuss@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/rdkit-discuss >> >> >> >> >> >> _______________________________________________ >> Rdkit-discuss mailing list >> Rdkit-discuss@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/rdkit-discuss >> > > > -- > > ====================================================================== > > Dr Thomas Evangelidis > > Research Scientist > > IOCB - Institute of Organic Chemistry and Biochemistry of the Czech > Academy of Sciences <https://www.uochb.cz/web/structure/31.html?lang=en> > Prague, Czech Republic > & > CEITEC - Central European Institute of Technology <https://www.ceitec.eu/> > Brno, Czech Republic > > email: teva...@gmail.com > > website: https://sites.google.com/site/thomasevangelidishomepage/ > > > _______________________________________________ > Rdkit-discuss mailing list > Rdkit-discuss@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/rdkit-discuss >
_______________________________________________ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
