Data to design: ExFAB’s Integrated workflows for AI/ML-driven functional genomics

Data to design: ExFAB’s Integrated workflows for AI/ML-driven functional genomics at  ACS Spring 2026, Atlanta Georgia, Georgia World Conference Center

ExFAB, the NSF-funded BioFoundry for Extreme and Exceptional Microbes, is building the experimental–computational pipeline required for AI- and ML-enabled biology — from automated data generation to predictive model development. A central focus is creating high-quality, large-scale genotype–phenotype datasets using standardized pooled screens and automated workflows. ExFAB’s efforts in this space began with creating ML suitable datasets for species-specific CRISPR guide design models with a focus on non-conventional yeasts. A key outcome of this work was establishing dataset design rules optimized for ML and AI modeling. One of ExFAB’s goals is to enable high throughput and automated workflows for any microbe. In moving toward this goal, we expanded our efforts to computational approaches that enable the prediction of active guides across the fungal kingdom; our ALLEGRO model accomplishes this. Current efforts to expand modeling to any microbe are focused on phenotypic data generation using PhenoTypic, an image-analysis workflow that captures quantitative growth, morphology, and color traits from large microbial libraries, generating the curated datasets required for supervised learning with AI models. Together, these integrated experimental and computational capabilities enable ExFAB to develop reliable predictive models that accelerate functional genomics, genome editing, and strain engineering across diverse microbial systems.