Research roundup: New application of machine learning accelerates and improves the hunt for antimicrobial resistant bacteria
Roughly two million people contract an infection that is resistant to antibiotics—and about 20,000 people die of such an infection—each year in the United States. Antimicrobial resistance (AMR) is encoded in bacterial genes and can be identified through next-generation sequencing—a process that tells scientists the genetic information being carried in DNA segments. This might make diagnosing and tracking AMR more accurate and rapid, but methods for detecting AMR genes often fail to identify new variations in the genes, which could lead to problems in detecting emerging AMR genes. Noelle Noyes, DVM, PhD, assistant professor in the Department of Veterinary Population Medicine, is working to overcome this challenge by collaborating with researchers at Colorado State University, University of Florida, and Texas A&M University to identify a faster way of identifying AMR genes with novel variations. Their method relies on machine learning and can be used to identify many AMR genes in a mixed community of bacteria from the same sample at the same time. The program, called meta-MARC, is breaking new ground by identifying more information from a varied set of AMR genes in both human- and environment-derived samples—and it can do so without compromising accuracy or timeliness. The researchers say this functionality is essential to identifying new AMR genes in the ongoing fight to combat AMR.
Learn more about this research in the August 6 issue of Communications Biology.