Molecular Mechanisms of Antimicrobial Resistance from Machine Learning Augmented Enhanced Sampling

Antimicrobial resistance threatens our ability to treat previously curable infectious diseases and may soon become a global public health crisis. The Ray group aims to understand and characterize the molecular mechanisms of antibiotic and antiviral resistance to identify potential avenues to target resistant pathogens. This R35 MIRA pro- gram involves two distinct research projects that utilize advanced machine learning (ML) and enhanced sampling algorithms for molecular dynamics (MD) simulations to gain mechanistic insights into antimicrobial resistance and facilitate the development of future therapeutic applications. In the first project, we will study the process of ligand binding to riboswitches, a class of regulatory RNA segments that are potential targets for next-generation antibi- otics. Our goal is to identify the role of conformational dynamics and distant nucleotide mutations in modulating the binding mechanism of the small molecule inhibitors (e.g., Ribocil) to RNA targets (e.g., Flavin-mononucleotide (FMN) riboswitch). We will design neural network (NN) and explainable artificial intelligence (XAI) based collec- tive variables from system agnostic descriptor space and perform enhanced sampling simulations to compute the free energy landscape of riboswitch conformational transition and ligand binding. This work will provide key mechanistic insights into RNA-small-molecule interactions and pave the way for designing more resilient antibi- otics. In the second project, we will study how resistant mutations in the viral antigens, e.g., SARS-CoV-2 spike protein, affect the binding mechanism of neutralizing antibodies. Previous research in this area primarily focused on the antigen-antibody interface but often overlooked the long-range allosteric effect of antigen mutations on the antibody binding process. We will perform NN and XAI-guided enhanced sampling simulations to elucidate the mechanistic details of antigen-antibody recognition. In addition, we will trace the allosteric communication path- ways using mutual-information-based protein graph connectivity networks constructed for various intermediate configurations sampled from the association pathway. This work will open new avenues for the rational design of broad-spectrum monoclonal antibodies through the judicious strengthening of distant regions of the antibody structure that are less susceptible to epitope mutations. Project Number: 1R35GM162443-01 | Fiscal Year: 2026 | NIH Institute/Center: National Institute of General Medical Sciences (NIGMS) | Principal Investigator: Dhiman Ray | Institution: UNIVERSITY OF OREGON, EUGENE, OR | Award Amount: $401,285 | Activity Code: R35 | Study Section: Special Emphasis Panel[ZRG1 MBBC-J (55)] View on NIH RePORTER: https://reporter.nih.gov/project-details/11269405