Benchmarking the Widely Used Structure-based Binding Affinity Predictors on the Spike-ACE2 Deep Mutational Interaction Set

ABSTRACT

Since the start of COVID-19 pandemic, a huge effort has been devoted to understanding the Spike(SARS-CoV-2)-ACE2 recognition mechanism. As prominent examples, two deep mutational scanning studies traced the impact of all possible mutations/variants across the Spike-ACE2 interface. Expanding on this, we benchmark four widely used structure-based binding affinity predictors (FoldX, EvoEF1, MutaBind2, SSIPe) and two naïve predictors (HADDOCK, UEP) on the variant Spike-ACE2 deep mutational interaction set. Among these approaches, FoldX ranks first with a 64% success rate, followed by EvoEF1 with a 57% accuracy. Upon performing residue-based analyses, we reveal algorithmic biases, especially in ranking mutations with increasing/decreasing hydrophobicity/volume. We also show that the approaches using evolutionary-based terms in their scoring functions misclassify most mutations as binding depleting. These observations suggest plenty of room to improve the conventional affinity predictors for guessing the variant-induced binding profile changes of Spike-ACE2. To aid the improvement of the available approaches we provide our benchmarking data at https://github.com/CSB-KaracaLab/RBD-ACE2-MutBench