Engcyc

Charting the interoperability of microbial systems across scales of biological organization.

Working Group: Pathways

A vast reservoir of functional dark matter persists within genomics, as experimental characterization has been achieved for only a minuscule fraction of the enzymes underlying life's chemical reactions. While new sequencing technologies and bioinformatic tools continue to accelerate data generation and analysis, the computational infrastructure needed to run these tools at scale remains lacking, and bespoke solutions are neither extensible long-term nor accessible to non-domain experts, leaving a conservative 34% of unique protein sequences unlinked to any known function. This project proposes a three-step AI framework to close that gap: large language models will integrate new bioinformatics tools into existing analysis infrastructure, facilitating adoption of state-of-the-art deep learning methods for functional prediction; agentic AI will lower accessibility barriers through a natural language interface layered atop a hallucination-free core system; and a unified analysis framework will enable automated chaining of tools into Nextflow-executable workflows. The resulting platform would provide an end-to-end bioinformatics solution for discovering novel enzymes, modeling microbial ecosystems, and engineering biological systems for sustainable bioproduction.