Phylogeny-agnostic ML predicts strain-level phage–host interactions
Researchers developed a phylogeny‑agnostic machine‑learning framework that predicts strain‑level phage–host interactions from genome sequences alone. The team evaluated over 13.2 million training runs across six datasets, covering 115,037 interactions, 949 bacterial strains and 518 phages. The workflow optimizes genome representations, feature selection and model choice to handle high…
Key points
- 13.2 million training runs on 115,037 interactions across 949 strains and 518 phages.
- Cross‑validation AUROC 0.67–0.94; experimental validation AUROC 0.84, 68.6 % mediators captured.
- Model‑guided cocktails reach 97.5 % coverage with five phages, 3.1‑fold better than promiscuity selection.
Cross‑validation shows AUROC ranging from 0.67 to 0.94, matching species‑specific methods. Experimental validation of 1,240 predicted E. coli phage–host pairs confirmed an AUROC of 0.84, and RB‑TnSeq screens verified that 68.6 % of identified infection mediators were captured computationally. Model‑guided cocktail design achieved up to 97.5 % bacterial coverage with five phages and a 3.1‑fold improvement over promiscuity‑based single‑phage selection.
The platform enables rational phage‑therapy design and precision microbiome engineering across clinical, agricultural and industrial contexts. By removing phylogenetic constraints, it expands the usable phage bank and reduces experimental burden for clinicians and researchers seeking strain‑specific phage candidates.
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- Researchers test 72,000 RAG combos on Indian government documents · 1 src
- Researchers release ChestPheNoT for auditable radiology report analysis · 1 src
- BioDyad integrates biomedical discovery with ML program search · 1 src
- Researchers propose MedCode to boost LLMs’ medical calculation accuracy by 20–30% · 1 src
- Researchers release open protocol for distributed AI textbook formalization · 1 src
Comments
via GitHub Discussions