International journal of antimicrobial agents · 2025
Machine-learning and gene-synergy networks reveal interpretable drivers of antibiotic resistance in Staphylococcus aureus.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Xu Jie, Li Yuan, Wang Ying, Liu Fang, Long Jinzhao, Zhu Jingyuan, Jin Yuefei, Chen Shuaiyin, Duan Guangcai, Yang Haiyan
Аннотация
INTRODUCTION: The evolution of antimicrobial resistance in Staphylococcus aureus (S. aureus) involves complex genotype-phenotype interactions of multiple genetic determinants acting independently.
OBJECTIVE: In this study, we aimed to accurately predict antimicrobial resistance from genomic sequences and uncover the complex genetic interactions underlying its mechanisms, focusing on interpretability and mechanistic insights.
METHODS: We propose a framework that elucidates antibiotic resistance by linking the genomic context with underlying gene interactions, combining a reference-agnostic gene-context 22-mer (gkmer) representation, a two-step random forest pipeline (RF1 screening/mapping; RF2 gene-level modelling), co-information-based synergy networks, and protein-structure mapping.
RESULTS: Using genomic sequences from 4569 S. aureus strains, we successfully predicted resistance to 12 antibiotics and examined the genetic interaction networks for penicillin, ciprofloxacin, and erythromycin in detail. On the held-out 20% test set, RF2 achieved an area under the receiver operating characteristic curve (AUC) of 0.83-1.00 (median 0.90) and an F1 score of 0.58-0.99 (median 0.76); comparable performance was observed for RF1. External validation on an independent dataset including 1011 S. aureus isolates revealed strong generalization for ciprofloxacin, erythromycin, and penicillin (area under the receiver operating characteristic curve > 0.95; F1 > 0.97), while highlighting reduced performance for clindamycin and tetracycline, and failure for gentamicin and trimethoprim, thereby delineating the scope and limits of the model and its applicability. Network analysis revealed distinct structural patterns of gene cooperation, highlighting key hubs and community structures consistent with known resistance pathways.
CONCLUSIONS: Our findings indicate that antibiotic resistance in S. aureus involves distinct genetic network architectures depending on the antibiotic, highlighting the diverse regulatory pathways bacteria employ to acquire resistance. This discovery-oriented framework prioritizes interpretable determinants and generates testable mechanistic hypotheses, with prospective standardized evaluations aimed at assessing its translatability to clinical settings.
OBJECTIVE: In this study, we aimed to accurately predict antimicrobial resistance from genomic sequences and uncover the complex genetic interactions underlying its mechanisms, focusing on interpretability and mechanistic insights.
METHODS: We propose a framework that elucidates antibiotic resistance by linking the genomic context with underlying gene interactions, combining a reference-agnostic gene-context 22-mer (gkmer) representation, a two-step random forest pipeline (RF1 screening/mapping; RF2 gene-level modelling), co-information-based synergy networks, and protein-structure mapping.
RESULTS: Using genomic sequences from 4569 S. aureus strains, we successfully predicted resistance to 12 antibiotics and examined the genetic interaction networks for penicillin, ciprofloxacin, and erythromycin in detail. On the held-out 20% test set, RF2 achieved an area under the receiver operating characteristic curve (AUC) of 0.83-1.00 (median 0.90) and an F1 score of 0.58-0.99 (median 0.76); comparable performance was observed for RF1. External validation on an independent dataset including 1011 S. aureus isolates revealed strong generalization for ciprofloxacin, erythromycin, and penicillin (area under the receiver operating characteristic curve > 0.95; F1 > 0.97), while highlighting reduced performance for clindamycin and tetracycline, and failure for gentamicin and trimethoprim, thereby delineating the scope and limits of the model and its applicability. Network analysis revealed distinct structural patterns of gene cooperation, highlighting key hubs and community structures consistent with known resistance pathways.
CONCLUSIONS: Our findings indicate that antibiotic resistance in S. aureus involves distinct genetic network architectures depending on the antibiotic, highlighting the diverse regulatory pathways bacteria employ to acquire resistance. This discovery-oriented framework prioritizes interpretable determinants and generates testable mechanistic hypotheses, with prospective standardized evaluations aimed at assessing its translatability to clinical settings.