International journal of antimicrobial agents · 2025
Genomic insights into nontyphoidal Salmonella: Prediction of antimicrobial resistance with whole genome-based machine learning.
СтатьяНазвание и аннотация: Перевод в очереди
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.
Woh Pei Yee, Soengkono Fadjar, Chen Yehao, Azizul Hasan Zati Hakim, Mohd Zain Siti Nursheena, Quiroga Jose, Kwok Kevin Wing Hin
Аннотация
BACKGROUND: Nontyphoidal Salmonella is a world-leading foodborne pathogen associated with an increased rate of antimicrobial resistance (AMR) and remains endemic in Asia. Utilizing whole genome sequencing (WGS) could significantly contribute to AMR prediction, from bioinformatic phylogenomic analysis to the advancement of machine learning (ML), leading towards automated AMR diagnostic.
METHODS: We obtained the Salmonella WGS from the National Centre for Biotechnology Information database and analysed their resistance profiles. We extracted, transformed, and labelled the resistance data with one-hot encoding platform for eXtreme Gradient Boosting (XGBoost) and convolutional neural network (CNN) model construction, training, and evaluation.
RESULTS: We selected a total of 788 Salmonella isolates associated with resistance genotype and phenotype data. These isolates had high resistance to aminoglycoside, beta-lactam, phenicol, quinolone, sulphonamide, tetracycline, and trimethoprim. S. Weltevreden ST365 (n = 121) was the most common serovar with the highest occurrence in food products. Through ML, both XGBoost and CNN models enabled highly accurate AMR prediction with performance accuracy of 0.97625 and 0.9904, respectively. Moreover, the interpretation of Shapley Additive exPlanations values uncovers the most valuable genomic features and associated genes for each antimicrobial agent tested.
CONCLUSIONS: Our study provides new knowledge in demonstrating the AMR phylogeographical relatedness and AMR prediction through XGBoost and CNN with competitive performance. Hence, WGS-based ML prediction and its machine application could be promoted as a promising tool for AMR work in food safety and public health settings. VIDEO ABSTRACT.
METHODS: We obtained the Salmonella WGS from the National Centre for Biotechnology Information database and analysed their resistance profiles. We extracted, transformed, and labelled the resistance data with one-hot encoding platform for eXtreme Gradient Boosting (XGBoost) and convolutional neural network (CNN) model construction, training, and evaluation.
RESULTS: We selected a total of 788 Salmonella isolates associated with resistance genotype and phenotype data. These isolates had high resistance to aminoglycoside, beta-lactam, phenicol, quinolone, sulphonamide, tetracycline, and trimethoprim. S. Weltevreden ST365 (n = 121) was the most common serovar with the highest occurrence in food products. Through ML, both XGBoost and CNN models enabled highly accurate AMR prediction with performance accuracy of 0.97625 and 0.9904, respectively. Moreover, the interpretation of Shapley Additive exPlanations values uncovers the most valuable genomic features and associated genes for each antimicrobial agent tested.
CONCLUSIONS: Our study provides new knowledge in demonstrating the AMR phylogeographical relatedness and AMR prediction through XGBoost and CNN with competitive performance. Hence, WGS-based ML prediction and its machine application could be promoted as a promising tool for AMR work in food safety and public health settings. VIDEO ABSTRACT.