Using generative AI to understand the biology of persistent Tuberculosis infection and improve treatment

Project Team: Gordon Broderick, Jeffrey Chen, Steven Rayan, Pranta Saha

Mycobacterium tuberculosis (Mtb) is the causative agent of tuberculosis (TB) which is resurging globally post- COVID-19 and disproportionately affects indigenous communities. Successful infection requires that M.tb actively deceiving the host cell’s natural defenses, particularly autophagy, a natural process in which cells break down unneeded or even dangerous components like bacteria or viruses.  How exactly Mtb carried out this deception is unclear.  This project hopes to decipher the molecular mechanisms that Mtb hijacks by assembling into a computer model what we currently know about the host cell’s molecular machinery using the latest advances in computerized natural language processing and database mining to construct a first skeleton model of the cell’s biochemical signaling. As our current knowledge continues to grow and is not complete, we will attempt to fill in the gaps in this initial model of cellular machinery by generating well-informed estimates of what might be missing using latest advances in generative artificial intelligence (AI).  Importantly we will guide our generative AI models by providing them with a foundation in cellular biochemistry to give context to our questions.  We will also teach them using specific examples from better-know components of immune cell function. By drawing on (i) what we know and (ii) what we believe might be true, we will use these large computer models of the molecular machinery to identify processes and specific components in the cell that are manipulated by Mtb during infection and design actionable therapeutic strategies for improving resistance to acute and persistent infection.