mahendra-awale / medchem_moves Goto Github PK

Hi and thanks for this inspiring work!

I am able to reproduce the example in the readme file - but unfortunately not able to run the compound generator with my compounds of choice. I repeatedly get the error: "Replace group XXX not found in fragmentation of input molecule"
I have initially pasted a fragment SMILEs from ChemDraw - which didn't work, then have fragmented my compound using the 'mmpdb fragment' command and have extracted several fragment SMILES from the generated text file - but with the same results. Here is one of my run command (with a fragment pasted from the mmpdb fragment output):

python .\mmpdb mmpCompoundGenerator --ttimes --tjobs 4 --transformation_db chemblDB3.sqlitdb --tmin-pairs 100 --toutput CFN92305_transforms.tsv --tsmiles "CC(C)[C@H]1CC[C@@]2(C)[C@@H]3[C@@H](O[C@@H]4O[C@H](CO)[C@@H](O)[C@H](O)[C@H]4O)C=C4[C@@H](CC[C@H](O)C4(C)C)[C@]3(C=O)CC[C@]12C" --replaceGroup "*C/C=C/C(C)(C)OC"

At the same time, I would like to ask if it's possible to run the compound generator on a file containing several compounds, similar to the way 'mmpdb transform' works?

https://figshare.com/articles/dataset/chemblDB3_sqlitdb/12912080

Hi the ChEMBL moves are under permanent embargo on figshare. Is there another place we can get them from?

Hi authors,

First of all, it has been a great work on using the MMP-DB as a powerful generator.

I have a question around how to generate the MMP_database that is compatible with the generator. I tried the old mmpdb code (https://github.com/rdkit/mmpdb) to generate the database (fragment + indexing), but I can't use that with mmpCompoundGenerator command (sqlite3.OperationalError: no such table: fragment_smi). Would you mind give an example of how to generate the MMP_DB for the generator? That would be very helpful.

Also, in your paper Figure S1, you mentioned new fragmentation and indexing steps have been implemented to enable parallel computing. It would be very useful for large datasets. I wonder if the new fragmentation and indexing functions are also available?

Thanks
Cheng

Hi Mahendra
I was doing some experiments to see if our 'playbook'@Novartis was similar, but it appears that index_algorithm.py is missing, and cannot be directly replaced by the standard mmpdblib version as the number of arguments is not the same. I think the rest of the MANIFEST is there. I therefore cannot see why norm_smi_file is needed, nor run mmpdb index foo.fragments