Skip to content
DraconiticResearch
Search Sources API
203,203 studies31 sourcesUpdated 27 min ago
Peer-reviewedCC BY-NC-ND 4.0Nature Microbiology

Phylogeny-agnostic strain-level prediction of phage–host interactions from genomes using machine learning

  • Avery J. C. Noonan1
  • Lucas Moriniere1,2
  • Edwin O. Rivera-López3
  • Krish Patel4
  • Melina Pena1
  • Madeline Svab1,5,6
  • Alexey Kazakov1
  • Adam Deutschbauer1,6
  • Edward G. Dudley3,7
  • Vivek K. Mutalik2
  • Adam P. Arkin1,2,4
7 affiliations
  1. 1Environmental Genomics and Systems Biology, Lawrence Berkeley National Laboratory, Berkeley, CA USA
  2. 2California Institute for Quantitative Biosciences, University of California Berkeley, Berkeley, CA USA
  3. 3Department of Food Science, The Pennsylvania State University, University Park, PA USA
  4. 4Department of Bioengineering, University of California Berkeley, Berkeley, CA USA
  5. 5Department of Molecular and Cell Biology, University of California Berkeley, Berkeley, CA USA
  6. 6Department of Plant and Microbial Biology, University of California Berkeley, Berkeley, CA USA
  7. 7E. coli Reference Center, The Pennsylvania State University, University Park, PA USA
Read full textPDFDOICite

Abstract

Bacteriophages offer promising alternatives to antibiotics for treating drug-resistant infections and engineering microbiomes, but applications are limited by challenges related to selection of phages infecting specific bacterial strains. Here we present a phylogeny-agnostic machine-learning framework predicting strain-level phage–host interactions across diverse bacterial genera from genome sequences alone. Systematically optimizing the workflow over 13.2 million training runs across six datasets (115,037 interactions, 949 bacterial strains, 518 phages), we achieved performance matching species-specific methods (AUROC 0.67–0.94) while eliminating phylogenetic constraints. Experimental validation of 1,240 predicted E. coli phage–host interactions confirmed generalizability (AUROC 0.84), while genome-wide RB-TnSeq screens verified that 68.6% of experimentally identified infection mediators were captured computationally. Model-guided cocktail design achieved up to 97.5% bacterial coverage with five phages, and up to a 3.1-fold improvement in single-phage selection over promiscuity-based selection. This platform enables rational phage-therapy design and precision microbiome engineering with applications across clinical, agricultural and industrial contexts.

Related

  1. MarkerScout: A Disease-Agnostic Machine Learning Framework for Biomarker Prediction from Multi-Scale Mechanistic Models
    Robert Moore, Frank Agayie-Ntim, Lindsey B. Crawford +5
    arXivPreprintCC BY-SA 4.0Full textGeneral biologyarXiv
  2. Pathogen-mediated selection and phylogeny shape MHC architecture in 33 bat genomes
    Luke W. Silver, Dominik W. Melville, Andrea L. Schraven +5
    Frontiers in ImmunologyPeer-reviewedCC BY 4.0Full textEcologyCrossrefEurope PMCEurope PMCPMC
  3. Reference genomes and fossils revise bat family phylogeny and biogeography
    Ariadna Morales, Yan Liang, William Thomas +133
    NaturePeer-reviewedCC BY 4.0Full text2 citedEcologyHAL
  4. Ancient human genomes reveal farmer-pastoralist interactions in the formation of Iron Age Qin people
    Xirui Wang, Rui Wang, Hao Ma +17
    iSciencePeer-reviewedCC BY-NC 4.0Evolutionary biologyHuman geneticsCell PressCrossref
  5. EpiRank: partner-agnostic surface-patch retrieval for epitope prediction of protein binding interfaces
    B. Lin, D. Lu
    bioRxivPreprintCC BY-NCFull textBioinformatics & computational biologybioRxiv

Cite

BibTeXRIS
BibTeX
@article{noonan2026phylogenyagnostic,
  title = {{Phylogeny-agnostic strain-level prediction of phage–host interactions from genomes using machine learning}},
  author = {Avery J. C. Noonan and Lucas Moriniere and Edwin O. Rivera-López and Krish Patel and Melina Pena and Madeline Svab and Alexey Kazakov and Adam Deutschbauer and Edward G. Dudley and Vivek K. Mutalik and Adam P. Arkin},
  journal = {Nature Microbiology},
  year = {2026},
  month = sep,
  doi = {10.1038/s41564-026-02482-5},
  url = {https://doi.org/10.1038/s41564-026-02482-5},
  note = {CC BY-NC-ND 4.0},
}
RIS
TY  - JOUR
TI  - Phylogeny-agnostic strain-level prediction of phage–host interactions from genomes using machine learning
AU  - Avery J. C. Noonan
AU  - Lucas Moriniere
AU  - Edwin O. Rivera-López
AU  - Krish Patel
AU  - Melina Pena
AU  - Madeline Svab
AU  - Alexey Kazakov
AU  - Adam Deutschbauer
AU  - Edward G. Dudley
AU  - Vivek K. Mutalik
AU  - Adam P. Arkin
PY  - 2026
DA  - 2026/09/29
JO  - Nature Microbiology
DO  - 10.1038/s41564-026-02482-5
UR  - https://doi.org/10.1038/s41564-026-02482-5
AB  - Bacteriophages offer promising alternatives to antibiotics for treating drug-resistant infections and engineering microbiomes, but applications are limited by challenges related to selection of phages infecting specific bacterial strains. Here we present a phylogeny-agnostic machine-learning framework predicting strain-level phage–host interactions across diverse bacterial genera from genome sequences alone. Systematically optimizing the workflow over 13.2 million training runs across six datasets (115,037 interactions, 949 bacterial strains, 518 phages), we achieved performance matching species-specific methods (AUROC 0.67–0.94) while eliminating phylogenetic constraints. Experimental validation of 1,240 predicted E. coli phage–host interactions confirmed generalizability (AUROC 0.84), while genome-wide RB-TnSeq screens verified that 68.6% of experimentally identified infection mediators were captured computationally. Model-guided cocktail design achieved up to 97.5% bacterial coverage with five phages, and up to a 3.1-fold improvement in single-phage selection over promiscuity-based selection. This platform enables rational phage-therapy design and precision microbiome engineering with applications across clinical, agricultural and industrial contexts.
SN  - 2058-5276
ER  -