Hi Leon, Any existing conformers of the molecule are ignored when generating new conformers, so as long as the atom ordering is the same and you use the same random-number seed you should get the same conformers whether you start from 3D coordinates, 2D coordinates, or no coordinates.
-greg On Tue, Aug 3, 2021 at 6:26 AM topgunhaides <sunzhi....@gmail.com> wrote: > Hi Greg, > > Thanks a lot for the explanation. Sorry for the late response as I > was doing more tests. > So I guess that the SDF (with 3D coordinate) and SDF (with only 2D > coordinates) would also give me the same results, as long as both use the > same atom ordering and random seed numbers? > To put it another way, does the 3D coordinate in SDF input give any > advantages over the 2D SDF, in producing conformers with higher quality? Or > are the results also only atom-order and random seed dependent? > > I was using SDF with experimental 3D coordinates as input to generate > conformers and reproduce that experimental conformation. Although RDKit's > ETKDG is stochastic, I want to make sure that the input with some > predefined information (like experimental 3D coordinates in SDF or stereo > info in SMILES sting) will not affect the results or give biased results. > Thank you! > > Best, > Leon > > > On Tue, Jul 27, 2021 at 12:59 AM Greg Landrum <greg.land...@gmail.com> > wrote: > >> Hi Leon, >> >> The results of EmbedMultipleConfs() have a random component and are >> atom-order dependent. The first of these will cause you to get different >> results across runs even if you use exactly the same molecule unless you >> set the random number seed. >> The atom-order dependance means that if the SMILES and SDF have different >> atom orderings you would get different conformers even if you set the same >> random number seed. If the atom orderings are the same, however, you should >> get the same results. >> >> I hope this helps, >> -greg >> >> >> On Mon, Jul 26, 2021 at 5:04 PM topgunhaides <sunzhi....@gmail.com> >> wrote: >> >>> Hi guys, >>> >>> By calling "EmbedMultipleConfs", are different results expected by >>> changing from one input format to another? Like change it from a SDF (with >>> 3D coordinates) to a SMILES string? >>> Thank you! >>> >>> Leon >>> >>> _______________________________________________ >>> 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
