Corresponding Projections for Orphan Screening
This addresses a key bottleneck in drug discovery by enabling predictions for proteins without experimental data, though it appears to be an incremental improvement over existing transfer learning methods.
The paper tackles the problem of predicting binding affinities of compounds to orphan proteins (with no training data) by proposing a transfer learning approach called corresponding projections, which constructs models for orphan proteins from existing protein models while preserving protein similarity relationships. The authors report their method outperforms state-of-the-art approaches in orphan screening.
We propose a novel transfer learning approach for orphan screening called corresponding projections. In orphan screening the learning task is to predict the binding affinities of compounds to an orphan protein, i.e., one for which no training data is available. The identification of compounds with high affinity is a central concern in medicine since it can be used for drug discovery and design. Given a set of prediction models for proteins with labelled training data and a similarity between the proteins, corresponding projections constructs a model for the orphan protein from them such that the similarity between models resembles the one between proteins. Under the assumption that the similarity resemblance holds, we derive an efficient algorithm for kernel methods. We empirically show that the approach outperforms the state-of-the-art in orphan screening.