Predicting Antimicrobial Activity for Untested Peptide-Based Drugs Using Collaborative Filtering and Link Prediction

Abstract

The increase of bacterial resistance to currently available antibiotics has underlined the urgent need to develop new antibiotic drugs. Antimicrobial peptides (AMPs), alone or in combination with other peptides and/or existing antibiotics, have emerged as promising candidates for this task. However, given that there are thousands of known AMPs and an even larger number can be synthesized, it is inefficient to comprehensively test all of them using standard wet lab experimental methods. These observations stimulated an application of machine-learning methods to identify promising AMPs. Currently, machine learning studies frequently combine very different bacteria without considering bacteria-specific features or interactions with AMPs. In addition, the sparsity of current AMP data sets of antimicrobial activity disqualifies the application of traditional machine-learning methods or renders the results unreliable. Here we present a new approach, featuring neighborhood-based collaborative filtering, to predict with high accuracy a given bacterias response to untested AMPs, AMP-AMP combinations, and AMP-antibiotic combinations based on similarities between bacterial responses. Furthermore, we also developed a complementary bacteria-specific link approach that can be used to visualize networks of AMP-antibiotic combinations, enabling us to suggest new combinations that are likely to be effective. Our theoretical analysis of AMP physico-chemical features suggests that there is an optimal similarity between two different AMPs that exhibit strong synergistic behavior. This principle, alongside with our specific results, can be applied to find or design effective AMP-AMP combinations that are target-specific. Author summaryIt is well known that combinations of different antimicrobial peptides (AMPs), in comparison to single AMP species, can lead to more efficient antimicrobial activity, but the large number of possible combinations requires the application of efficient machine-learning methods. We developed a new approach consisting of collaborative filtering, link prediction, and AMP feature analysis to predict previously-unknown, bacteria-specific activity of AMP combinations, suggest novel synergistic AMP-antibiotic combinations, and guide future design of effective AMP-AMP combinations.