Back

Yuel: Compound-Protein Interaction Prediction with High Generalizability

Wang, J.; Dokholyan, N. V.

2021-07-08 bioinformatics
10.1101/2021.07.06.451043 bioRxiv
Show abstract

Virtual drug screening has the potential to revolutionize the stagnant drug discovery field due to its low cost and fast speed. Predicting binding affinities between small molecules and the protein target is at the core of computational drug screening. Deep learning-based approaches have recently been adapted to predict binding affinities and claim to achieve high prediction accuracy in their tests, however, we show that current approaches are not reliable for virtual drug screening due to the lack of generalizability, i.e. the ability to predict interactions between unknown proteins and unknown small molecules. To address this shortcoming, we develop a compound-protein interaction predictor, Yuel. Upon comprehensive tests on various datasets, we find that out of all the deep-learning approaches surveyed, only Yuel can predict interactions between unknown compounds and unknown proteins. Additionally, Yuel can also be utilized to identify compound atoms and proteins residues that are involved in the binding.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.