Hello, RDKit is getting a lot of traction these days because big corps have decided to put their weight behind it. My hunch is that they are choosing RDKit because it has put a lot of efforts in the recent years in their Python wrappers. And you know, Python and Deep Learning. A big deal. They also have a Java wrapper that they maintain for the KNIME platform. And they have Neo4J and Lucene toolkits, but Maven packages are nowhere to be found. We do ML in Python here too, of course, but all of our ETL is all in open source Java (Kafka, Hadoop, Neo4J, Lucene, Streamsets, NiFi, CoreNLP, etc.). And to me, a chemistry toolkit fits in the ETL (beginning of the pipeline) and inference (end of the pipeline) stacks, not in the offline ML training (mid-pipeline) stack of a system. And I'm not going to talk about Dragon, because it's a closed-source, proprietary software. So, I'd like to stick to CDK for the time being and find the features I'm looking for in it.
In particular, I have been looking for the famous 21 Dragon olfactive descriptors (also implemented for the most part in RDKit, I think). I have looked in the CDK Javadoc and source code and while I have found some of the keywords, I'm not sure they are the same or even remotely close. So could someone with more experience than I with CDK tell me if I can generate those descriptors from a SMILES-derived IAtomContainer in CDK ? And if I can be of any help, I'd be happy to contribute in implementing the missing ones. Thanks! Contrebande
_______________________________________________ Cdk-user mailing list Cdk-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/cdk-user
