Is it possible to skip step 1? I am not interested in using molecular fingerprinting and would like to skip step 1 to save time.
Created by Igor Tetko itetko Thank you for sharing this information! We received 116 valid submissions from 49 different teams/participants.
Considering only the best results, 39 groups were able to identify at least one active molecule in the selected 200 molecules, and 41 groups identified at least one active molecule in the selected 500. 21 groups identified at least three different clusters in their 200 molecules, and 35 in their selected 500.
Regardless of the results all groups are invited to participate in step 2. The aim is to compare the performance of the fingerprints-only models with other approaches which can use both the DEL-data and other hit discovery methods (ML models based on SMILES, docking, pharmacophores). About 140 teams... How many teams submitted (valid) predictions and thus advanced to the 2nd step? The training data set is a partnership with HitGen. They prefer not to disclose SMILES. Actually, even academics prefer not to disclose the SMILES of their DEL libraries. The good news is that the SGC's next generation of DEL data will include SMILES. This will take a few months. What is the reasoning behind not providing SMILES for the training data in Step 2? I think the differences in modelling results between Steps 1 and 2 will be limited if the training data remains identical. Step 1 is mandatory. We are interested in comparing step1 and step2 results. A reminder that step2 will provide SMILES for the test set but not for the training set.
Drop files to upload
Is participation in Step 1 mandatory? page is loading…