Assessing computational predictions of antimicrobial resistance phenotypes from microbial genomes

Description

This dataset contains the following materials: 1. ZIP file of txt files. PATRIC Genome IDs and AMR phenotypes, with '1' indicating resistance and '0' indicating susceptibility, for 78 single-species-antibiotic datasets. 2. ZIP file of JSON files. 223 (78+67+78) sets of random folds, phylogeny-aware folds, and homology-aware folds for 78 single-species-antibiotic datasets. Nine sets of homology-aware folds for nine multi-antibiotic datasets, each corresponding to one of nine species. A set of homology-aware folds for evaluating Aytan-Aktug control multi-species model and cross-species model. Nine sets of homology-aware folds for evaluating the Aytan-Aktug leave-one-species-out cross-species model. 3. ZIP file of PDF files. Phylogenetic trees generated by Geno2Pheno for 10 species (including 67 species-antibiotic combinations), annotated with phenotypes and the cross-validation folds for phylogeny-aware and random folds, respectively.

Steps to reproduce

https://github.com/hzi-bifo/AMR_benchmarking

Categories

Funding

Licence