The World Health Organization describes antimicrobial resistance (AMR) as an increasingly serious threat to global public health. This threat has prompted the President of the United States to issue an executive order initiating the National Action Plan for Combating Antibiotic-Resistant Bacteria. The fifth goal of this Action Plan is to: "Improve International Collaboration and Capacities for Antibiotic-resistance Prevention, Surveillance, Control, and Antibiotic Research and Development." This plan also recognizes that humans, animals, and the environment can be sources of AMR.
Despite the growing concern with antibiotic use in agriculture, there are very few computational methods to measure, quantify, and track the AMR genes within a food production system. Hence, one of my research interests is in the development of methods that will allow for systematic surveillance of the microbiome, or more specifically the resistome, within a food production system and its' surrounding environment. The resistome refers to the collection of all antibiotic resistance genes, and their precursors, in both pathogenic and non-pathogenic bacteria. This surveillance is made possible by advances in our ability to perform metagenomics sequencing to survey food production systems (specifically meat production systems), and agricultural environments, at a large scale and at frequent time intervals. In order to achieve full-scale surveillance framework, my research lab has been tackling the following research problems.
AMR gene identification and quantification. We have completed preliminary analysis to determine the diversity of AMR genes contained in a sample using a database we have compiled containing approximately 4,000 unique AMR gene sequences. Current methods for AMR gene identification and quantification require assembly of the metagenomic data followed by translation of the assembled regions into protei n sequences; however, both processes are highly error-prone. We are developing a novel approach that avoids both assembly and translation of the data---thus increasing the accuracy of AMR gene identification. In addition, to the identification of AMR genes, this method will quantification of the abundance of AMR genes within observed samples, resulting in an AMR-Gene profile per sample.
Variant profile identification and quantification. My lab is building more targeted methods to identify and quantify the rare single nucleotide polymorphism (SNP) or indel (insertion and/or deletion) profiles for our database of AMR genes. Our preliminary analysis uses read alignment in order to accomplish this task but there are shortcomings to this analysis that affect the accuracy of the estimation. In addition to increasing the accuracy of detection, our method will use succinct data structures to ensure that a large number of samples can be analyzed simultaneously. %Determine the dissemination pathway of the mcrobiome and resistome between the feedlot and environment. To accomplish this we will develop a resistance gene SNP database and variant calling pipeline to identify rare SNP or indel (insertions or deletions) profiles that act as ``fingerprints'' in determining the source of AMR genes.
AMR-gene comparison and variant profile classification. Lastly, we plan to develop a statistical analysis framework that compares and contrasts AMR-Gene profiles between different samples, food production systems, and/or processing stages (using the output of the methods developed for the first problem), and utilizes validated and predictive sample variant profiles to classify new samples in order to identify their origin (using the output of the methods developed for the second problem). This project will be done in collaboration with specialists in epidemiology.
Together these projects will provide an efficient framework to accurately characterize AMR genes in metagenomic samples, and to identify their likely origin. Our methods could potentially have a huge impact on food safety and agricultural practices. For instance, if highly resistant strains of bacteria are found in various locations in a food production system, and these strains are associated with specific antibiotic use or other production practices, then policy guidance could be made.
The Boucher lab greatfully receives funding from USDA NIFA for this research.