OpenAI profiles a bioengineering lab using Codex and ChatGPT to hunt for new antimicrobial molecules

OpenAI has published a case study on how a University of Pennsylvania bioengineering lab uses Codex and ChatGPT alongside its own deep-learning models to search genomes for candidate antimicrobial molecules, as drug-resistant infections continue to outpace the discovery of new antibiotic classes.

What's new

The subject of the case study is Cesar de la Fuente, a bioengineer whose lab searches for antimicrobial candidates by treating biological sequences as an information system. "The nucleotides that make up DNA, and the amino acids that make up proteins and peptides are sort of like an alphabet," de la Fuente said. His lab trains deep-learning models to recognize patterns in genome and protein databases that correlate with antimicrobial activity, an approach OpenAI says "can reduce the initial search for candidate molecules from years to hours."

On top of the lab's own models, "the lab uses ChatGPT and Codex to brainstorm hypotheses, write and refine code, process datasets, analyze results, and connect ideas across scientific disciplines." De la Fuente describes the tools as a brainstorming partner and collaborative sounding board for lab members, including non-native English speakers, though he stresses the need to verify AI output: "Obviously you have to always double-check for accuracy."

Context

Drug-resistant bacteria, fungi, parasites, and viruses are a growing public-health threat; OpenAI cites an estimate that antimicrobial resistance was associated with roughly five million deaths in 2021, a toll projected to roughly double by 2050. De la Fuente notes the field has not produced a new class of antibiotics in 50 years, in part because most development modifies existing drug families rather than searching genuinely new chemical space. His lab's approach — mining largely unannotated regions of genomes and proteomes — is aimed at that unexplored territory, using AI to compress a search that traditionally took years of lab work.

Why it matters

The case study is a narrow, single-lab example rather than a product launch, but it is a concrete illustration of OpenAI positioning its general-purpose coding and chat tools as infrastructure for scientific discovery pipelines, alongside domain-specific models built by researchers themselves. For a field where the bottleneck is often narrowing a vast genomic search space rather than running a single experiment, tools that speed up hypothesis generation, data processing, and cross-disciplinary idea connection could shorten timelines in an area where antibiotic development has stalled for decades.