Inferring strain-level mutational drivers of phage-bacteria interaction phenotypes.

The enormous diversity of bacteriophages and their bacterial hosts presents a significant challenge to predict which phages infect a focal set of bacteria. Infection is largely determined by complementary -and largely uncharacterized- genetics of adsorption, injection, and cell take-over. Here we present a machine learning (ML) approach to predict phage-bacteria interactions trained on genome sequences of and phenotypic interactions amongst 51 Escherichia coli strains and 45 phage λ strains that coevolved in laboratory conditions for 37 days. Leveraging multiple inference strategies and without a priori knowledge of driver mutations, this framework predicts both who infects whom and the quantitative levels of infections across a suite of 2,295 potential interactions. The most effective ML approach inferred interaction phenotypes from independent contributions from phage and bacteria mutations, predicting phage host range with 86% mean classification accuracy while reducing the relative error in the estimated strength of the infection phenotype by 40%. Further, transparent feature selection in the predictive model revealed 18 of 176 phage λ and 6 of 18 E. coli mutations that have a significant influence on the outcome of phage-bacteria interactions, corroborating sites previously known to affect phage λ infections, as well as identifying mutations in genes of unknown function not previously shown to influence bacterial resistance. While the genetic variation studied was limited to a focal, coevolved phage-bacteria system, the method's success at recapitulating strain-level infection outcomes provides a path forward towards developing strategies for inferring interactions in non-model systems, including those of therapeutic significance.

Modeling the synergistic elimination of bacteria by phage and the innate immune system.

Phage therapy has been viewed as a potential treatment for bacterial infections for over a century. Yet, the year 2016 marks one of the first phase I/II human trials of a phage therapeutic - to treat burn wound patients in Europe. The slow progress in realizing clinical therapeutics is matched by a similar dearth in principled understanding of phage therapy. Theoretical models and in vitro experiments find that combining phage and bacteria often leads to coexistence of both phage and bacteria or phage elimination altogether. Both outcomes stand in contrast to the stated goals of phage therapy. A potential resolution to the gap between models, experiments, and therapeutic use of phage is the hypothesis that the combined effect of phage and host immune system can synergistically eliminate bacterial pathogens. Here, we propose a phage therapy model that considers the nonlinear dynamics arising from interactions between bacteria, phage and the host innate immune system. The model builds upon earlier efforts by incorporating a maximum capacity of the immune response and density-dependent immune evasion by bacteria. We analytically identify a synergistic regime in this model in which phage and the innate immune response jointly contribute to the elimination of the target bacteria. Crucially, we find that in this synergistic regime, neither phage alone nor the innate immune system alone can eliminate the bacteria. We confirm these findings using numerical simulations in biologically plausible scenarios. We utilize our numerical simulations to explore the synergistic effect and its significance for guiding the use of phage therapy in clinically relevant applications.

Conflicting attachment and the growth of bipartite networks.

Simple growth mechanisms have been proposed to explain the emergence of seemingly universal network structures. The widely studied model of preferential attachment assumes that new nodes are more likely to connect to highly connected nodes. Preferential attachment explains the emergence of scale-free degree distributions within complex networks. Yet it is incompatible with many network systems, particularly bipartite systems in which two distinct types of agents interact. For example, the addition of new links in a host-parasite system corresponds to the infection of hosts by parasites. Increasing connectivity is beneficial to a parasite and detrimental to a host. Therefore, the overall network connectivity is subject to conflicting pressures. Here we propose a stochastic network growth model of conflicting attachment, inspired by a particular kind of parasite-host interaction: that of viruses interacting with microbial hosts. The mechanism of network growth includes conflicting preferences to network density as well as costs involved in modifying the network connectivity according to these preferences. We find that the resulting networks exhibit realistic patterns commonly observed in empirical data, including the emergence of nestedness, modularity, and nested-modular structures that exhibit both properties. We study the role of conflicting interests in shaping network structure and assess opportunities to incorporate greater realism in linking growth process to pattern in systems governed by antagonistic and mutualistic interactions.