AI-Bind: Improving Binding Predictions for Novel Protein Targets and Ligands

Abstract

Identifying novel drug-target interactions (DTI) is a critical and rate limiting step in drug discovery. While deep learning models have been proposed to accelerate the identification process, we show that state-of-the-art models fail to generalize to novel (i.e., never-before-seen) structures. We first unveil the mechanisms responsible for this shortcoming, demonstrating how models rely on shortcuts that leverage the topology of the protein-ligand bipartite network, rather than learning the node features. Then, we introduce AI-Bind, a pipeline that combines network-based sampling strategies with unsupervised pre-training, allowing us to limit the annotation imbalance and improve binding predictions for novel proteins and ligands. We illustrate the value of AI-Bind by predicting drugs and natural compounds with binding affinity to SARS-CoV-2 viral proteins and the associated human proteins. We also validate these predictions via auto-docking simulations and comparison with recent experimental evidence. Overall, AI-Bind offers a powerful high-throughput approach to identify drug-target combinations, with the potential of becoming a powerful tool in drug discovery.

A new framework for evaluating model out-of-distribution for the biochemical domain

“…First, the population distribution is virtually infinite, whereas training distributions are generally small and significantly biased [23]. A common example is binding affinity databases being biased towards certain protein families and high affinity binders [7]. Secondly, individual data points are not independent from each other, as similar molecules share similar properties [44, 41] (e.g., two protein sequences with a sequence identity above 30% usually share a common ancestor and adopt similar structures [35]).…”