LibMCS options

Dear Miklos,





I run the LibMCS 0.5.1 on the attached dataset. I tried different parameters and expectedly came out with different top level MCS. I put all the results with different settings in the attached powerpoint file. The crucial point seems to be the minimal MCS size. Matching parameters (atom type/bond type) have no influence on the results which would give different results with the present dataset.


How would you deal with such data?


Is there a way to pre-optimize these different parameters?





Thank you very much.





Best regards,





ISI

Hi ISI,





the problem with the matching parameters was found, it is corrected in LibMCS 0.5.2 available at www.chemaxon.com/shared/libMCS.





Miklos is on holiday until 25 July, he will answer your questions as soon as he is back.





Kind regards,


Judit

Hi Isi,





I reckon the only issue to be answered is the pre-optimization of various parameters, options. We'll if you could suggest some viable strategy, we were happy to implement it! At present there is one similar option, the additive clustering that allows to gradually decrease the minimum MCS size / have you tried it or you need some further explanation on this feature?





Regards,


Miklos

I am looking forward to a class:


to generate at least


scaffolds out[0][] &


[R groups] i.e., out[1][]


for a given SDF/SMILES








The expected pseudo code is..





public String[][] GenerateScaffolds(String[] Smiles){





String[] scaff = new String[100];


String[] fgi = new String[100];


String[][] out = new String[2][100];


out[0][]= scaff;


out[1][]=fgi;





...





...





return out;





}





Thanks in advance

We will consider this and let you know how it fits in present API and our development plans.





Thanks for your suggestion.





Miklos