DebiasedDTA: Model Debiasing to Boost Drug-Target Affinity Prediction

Motivation: Computational models that accurately identify high-affinity protein-chemical pairs can accelerate drug discovery pipelines. These models, trained on available protein-chemical interaction datasets, can be used to predict the binding affinity of an input protein-chemical pair. However, the training datasets may contain surface patterns, called dataset biases, which cause models to memorize dataset-specific biomolecule properties, instead of learning binding mechanisms. As a result, the prediction performance of models drops for unseen biomolecules. Here, we present DebiasedDTA, a novel drug-target affinity (DTA) prediction model training framework that addresses dataset biases to improve affinity prediction for novel biomolecules. DebiasedDTA uses ensemble learning and sample weight adaptation to identify and avoid biases and is applicable to most DTA prediction models. Results: The results show that DebiasedDTA can boost models while predicting the interactions between unseen biomolecules. In addition, prediction performance for seen biomolecules also improves. The experiments also show that DebiasedDTA can augment DTA prediction models of different input and model structures and is able to avoid biases of different sources. The investigations of predictions reveal that model debiasing can diminish the importance of misleading features and can enable models to learn more from the proteins. DebiasedDTA is published as an open-source python package to enable debiasing custom DTA prediction models with only two lines of code.