Authors: Seshu Kumari K, Dr. Munesh Kumar Sharma
Abstract: The human gut microbiota is an established reservoir of antimicrobial resistance determinants, and sequence-based profiling of that reservoir is now routine. What remains unsolved is the step that follows: converting a list of resistance determinants into a statement expressed in the antimicrobial classes used at the point of prescribing. This paper specifies a six-stage computational framework that performs that conversion, and demonstrates it upon a panel of ten reference strains of the human colonic microbiota carrying six well-characterised determinants. The framework assigns determinants to drug classes of the Antibiotic Resistance Ontology, maps those onto sixteen clinical antimicrobial classes, applies an explicit guidance rule, and computes descriptive, ordination and exploratory analyses of the resulting strain-by-class matrix. All six stages executed to completion, producing 96 class-level decisions. Seven decisions asserting avoidance were generated, and each corresponds to an established resistance of the organism concerned: cepA-mediated cephalosporin resistance in both Bacteroides strains, AmpC and SHV β-lactamases in the Enterobacterales, tetW in Bifidobacterium longum, and the vanB operon and aac(6′)-Ii of Enterococcus faecalis V583. Transmission of input to output was correct in all nine testable cases. The determinant calls were curated from the primary genome literature rather than computed from sequence, so the demonstration establishes the behaviour of the mapping and guidance logic under correct input and does not evaluate any annotation tool. Within that limit, the panel design permits every intermediate value to be inspected by hand, and it is at this scale that the framework’s principal failure mode, reported separately, becomes visible.
DOI: https://zenodo.org/records/22977550
