MOBILE-PATH: Machine Learning Optimization for Lateral gene Exchange PATHways – Phase I
MOBILE-PATH (Machine-learning Optimization of Biological Interfaces for Lateral gene Exchange PATHways) will develop an AI framework to predict how mobile genetic elements (MGEs)—such as plasmids and other DNA carriers—move through microbial communities. Led by Princeton University, the project addresses a key barrier in microbiome engineering: the limited ability to anticipate which microbes exchange genetic material, how that exchange unfolds over time and how environmental conditions shape its spread.
The project will model horizontal gene transfer as movement through a changing network of microbial hosts. Using longitudinal data from the human microbiome and plant rhizosphere, the team will combine higher-order network models with causality-aware machine learning to predict multi-step MGE transfer pathways. It will also build an automated tool that uses metagenome-assembled genomes to estimate transfer likelihoods across different environments, drawing on predictions of the genetic machinery that enables transfer and the defenses that may prevent it.
MOBILE-PATH will include an agentic decision-making interface to help identify MGEs that could deliver a desired functional cargo to target microbes while limiting spread to non-target hosts. Together, these capabilities aim to make predictions of gene flow more generalizable across environments and microbiome functions.
The resulting framework could support more deliberate microbiome engineering, including delivery of useful genetic functions and evaluation of containment risks. By helping researchers screen candidate MGEs computationally, the project could reduce sequencing costs and the number of field trials needed to assess gene-delivery strategies.
Partners
- Princeton University
