We’ve stepped up recently to help out on JC Bradley’s CombiUgi Project. JC is working with the group at Indiana who have a service that predicts a compound’s activity in cell lines.

You can check out JC’s blog and Wiki about the ongoing work on CombiUgi…but what are we doing to help and how’s it going?

What we’ve done so far is deal with over 68,000 SMILES strings produced by Rajarshi and detailed here. Immediately we hit problems…a combination of SMILES conversion and InChI issues. During the initial population of the ChemSpider database we used a specific set of InChI layers. However, these have not matched up to the InChI searching algorithms implemented on ChemSpider in an optimal way. Specifically, issues with the Mobile Proton layer.

At present we are re-populating the database with newly generated InChi codes, have dealt with the SMILES string conversion issues and will be able to add and index the 68,000 CombiUgi products to the database shortly. The next piece is to work out with Rajarshi and JC how to expose the predictions associated with the cell lines and then later associate the measured data from the various cell lines. Watch this space…

Stumble it!

Leave a Reply