Inferring strain-level mutational drivers of phage-bacteria interaction phenotypes arising during coevolutionary dynamics DOI Creative Commons
Adriana Lucía-Sanz, Shengyun Peng, Chung Yin Leung

et al.

Virus Evolution, Journal Year: 2024, Volume and Issue: 10(1)

Published: Jan. 1, 2024

Abstract The enormous diversity of bacteriophages and their bacterial hosts presents a significant challenge to predict which phages infect focal set bacteria. Infection is largely determined by complementary—and uncharacterized—genetics adsorption, injection, cell take-over, lysis. Here we present machine learning approach phage–bacteria interactions trained on genome sequences phenotypic among 51 Escherichia coli strains 45 phage λ that coevolved in laboratory conditions for 37 days. Leveraging multiple inference strategies without priori knowledge driver mutations, this framework predicts both who infects whom the quantitative levels infections across suite 2,295 potential interactions. We found most effective inferred interaction phenotypes from independent contributions bacteria accurately predicting 86% while reducing relative error estimated strength infection phenotype 40%. Feature selection revealed key Escherchia mutations have influence outcome interactions, corroborating sites previously known affect infections, as well identifying genes unknown function not shown resistance. method’s success recapitulating strain-level outcomes arising during coevolutionary dynamics may also help inform generalized approaches imputing genetic drivers complex communities

Language: Английский

Making Waves: Intelligent phage cocktail design, a pathway to precise microbial control in water systems DOI
Bridget Hegarty

Water Research, Journal Year: 2024, Volume and Issue: 268, P. 122594 - 122594

Published: Oct. 9, 2024

Language: Английский

Citations

1
Inferring strain-level mutational drivers of phage-bacteria interaction phenotypes DOI Creative Commons
Adriana Lucía-Sanz, Shengyun Peng, Chung Yin Leung

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 9, 2024

Abstract The enormous diversity of bacteriophages and their bacterial hosts presents a significant challenge to predict which phages infect focal set bacteria. Infection is largely determined by complementary – uncharacterized genetics adsorption, injection, cell take-over lysis. Here we present machine learning approach phage-bacteria interactions trained on genome sequences phenotypic amongst 51 Escherichia coli strains 45 phage λ that coevolved in laboratory conditions for 37 days. Leveraging multiple inference strategies without priori knowledge driver mutations, this framework predicts both who infects whom the quantitative levels infections across suite 2,295 potential interactions. We found most effective inferred interaction phenotypes from independent contributions bacteria accurately predicting 86% while reducing relative error estimated strength infection phenotype 40%. Feature selection revealed key E. mutations have influence outcome interactions, corroborating sites previously known affect infections, as well identifying genes unknown function not shown resistance. method’s success recapitulating strain-level outcomes arising during coevolutionary dynamics may also help inform generalized approaches imputing genetic drivers complex communities

Language: Английский

Citations

0
Inferring strain-level mutational drivers of phage-bacteria interaction phenotypes arising during coevolutionary dynamics DOI Creative Commons
Adriana Lucía-Sanz, Shengyun Peng, Chung Yin Leung

et al.

Virus Evolution, Journal Year: 2024, Volume and Issue: 10(1)

Published: Jan. 1, 2024

Abstract The enormous diversity of bacteriophages and their bacterial hosts presents a significant challenge to predict which phages infect focal set bacteria. Infection is largely determined by complementary—and uncharacterized—genetics adsorption, injection, cell take-over, lysis. Here we present machine learning approach phage–bacteria interactions trained on genome sequences phenotypic among 51 Escherichia coli strains 45 phage λ that coevolved in laboratory conditions for 37 days. Leveraging multiple inference strategies without priori knowledge driver mutations, this framework predicts both who infects whom the quantitative levels infections across suite 2,295 potential interactions. We found most effective inferred interaction phenotypes from independent contributions bacteria accurately predicting 86% while reducing relative error estimated strength infection phenotype 40%. Feature selection revealed key Escherchia mutations have influence outcome interactions, corroborating sites previously known affect infections, as well identifying genes unknown function not shown resistance. method’s success recapitulating strain-level outcomes arising during coevolutionary dynamics may also help inform generalized approaches imputing genetic drivers complex communities

Language: Английский

Citations

0