search

rdkit / mmpdb

A package to identify matched molecular pairs and use them to predict property changes.
Other
198 stars 55 forks source link

How to output a list of ID of compounds forming pairs? #11

Closed den2882 closed 5 years ago

den2882 commented 5 years ago

This is not clear to me. Thanks

adalke commented 5 years ago

Do you mean when doing a prediction, as in the "mmpdb predict" described at https://github.com/rdkit/mmpdb#5-use-mmp-to-make-a-prediction ?

If so, the last paragraph of that section is:

You might try enabling the "--explain" option to see why the algorithm selected a given transformation, or use "--save-details" to save the list of possible rules to the file 'pred_detail_rules.txt' and to save the list of rule pairs to "pred_detail_pairs.txt".

Is that what you are looking for?

den2882 commented 5 years ago

I am using MMPs to navigate the SAR, which is a bit different from trying to compute statistics and rules for transformations. I would like to fragment the molecules using only typical synthetic routes (aryl-alkyl cuts and amides, or simply all sp3-sp2 bond cuts). Then retrieve the list of compounds forming pairs to review them manually.

For instance for 10 compounds with ID1,ID2,...,ID10. It would give me the following information: ID1 <--> ID3 ID1 <--> ID5 ID2 <--> ID4 etc...

adalke commented 5 years ago

An earlier version of mmpdb supported mmpdb rulecat to dump the rules from the database, but there wasn't a specific need or budget to update that for the public release.

What you can do instead is do mmpb index and specify a CSV file as the output. The mmpdb index --help output mentions "The 'csv' format is a tab-separated table with the columns: SMILES1 SMILES2 id1 id2 V1>>V2 C".

KramerChristian commented 5 years ago

Hi den2882,

does specifying the .csv output solve the issue? If yes, I would like to close it.

Best regards, Christian

den2882 commented 5 years ago

It worked. Thanks!