On 02/23/2015 06:12 AM, Markus Sitzmann wrote: > And I am not saying it is perfect, it just provides another > implementation to double-check things in question. It has the CACTVS > chemoinformatic toolkit as chemistry backend which I think is > well-tested.
> On Mon, Feb 23, 2015 at 10:54 AM, JP <jeanpaul.ebe...@inhibox.com> wrote: >> 3,257 molecules (of 6,940,083) gave me a different inchis between the >> current production version and the development (github) one. Just as another data point, out of 14 metabolites that's gone through my scripts so far: 1. cis-vaccenic acid, PubChem CID 5282761: InChI produced by RDKit 2014.09.2 differs from that from OpenBabel. InChI from PubChem's SDF agrees with the latter. (RDKit's ends with "/b8-7+", OB & PC: "7-".) InChI code does spit out "undefined stereochemistry" warnings for OB's "we don't need no C.I.P., it's sooo last century" stereo -- and then includes the stereo layer in the output anyway. (Though in this case PubChem's presumably OpenEye stereo seems to agree with OB and not RDKit.) 2. 5,10,15,20-Tetraphenyl-21H,23H-porphine zinc, PubChem CID 3580039: InChI in the SDF ends at the "/q", both RDKit and OpenBabel add "/b" layer. They all agree, though. 14 data points is nowhere near enough for meaningful conclusions, but still... 14% won't match the plain string comparison that most searches do and 7% won't match the "clever InChI-aware comparison" search, assuming it's implemented anywhere. And then you get .05% between different versions of the same software... -- Dimitri Maziuk Programmer/sysadmin BioMagResBank, UW-Madison -- http://www.bmrb.wisc.edu
signature.asc
Description: OpenPGP digital signature
------------------------------------------------------------------------------ Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server from Actuate! Instantly Supercharge Your Business Reports and Dashboards with Interactivity, Sharing, Native Excel Exports, App Integration & more Get technology previously reserved for billion-dollar corporations, FREE http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk
_______________________________________________ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
