ARSENAL: Antimicrobial ReSistance prEdictioN by a mAchine Learning method

Antimicrobial resistance (AMR) has emerged as a major public health concern due to the rapid proliferation of multidrug-resistant bacteria, posing significant challenges in the prevention and treatment of persistent infections. Developing algorithms for AMR prediction could be of great clinical importance, offering a more reliable and efficient alternative to traditional phenotyping methods and potentially contributing to the discovery of novel AMR pathways.
The recent increase in available sequencing data and associated phenotypic information has laid the foundation for the development of predictive methods. However, predicting antimicrobial resistance in terms of minimum inhibitory concentration (MIC) is challenging, as it requires more nuanced analysis than simply categorizing strains as susceptible or resistant. We have developed a machine learning approach called ARSENAL (Antimicrobial ReSistance prEdictioN by mAchine Learning) to predict the MIC of several antibiotics based on genomic data. ARSENAL utilizes single-nucleotide polymorphisms (SNPs) and takes into account the genome structure (gene composition) and gene orthology links between strains of the same species.
We have shown in the application to the analysis of ~1300 strains of Streptococcus pneumoniae that our ARSENAL model demonstrates high predictive accuracy in determining the minimum inhibitory concentration of various antibiotics. Functional interpretation of the most predictive features confirmed the biological relevance of the ARSENAL model, highlighting its capacity to identify key genetic determinants of AMR.
ARSENAL is a novel machine learning approach for predicting antimicrobial resistance based on genomic data. By leveraging SNPs, genome structure, and gene orthology information, ARSENAL offers a powerful tool for understanding and predicting AMR, with the potential to guide clinical decision-making and contribute to the development of novel therapeutic strategies to combat the growing threat of multidrug-resistant infections.