César de la Fuente’s lab uses OpenAI tools for coding, data preparation and hypothesis work while specialized models rank antimicrobial candidates.
Bioengineer César de la Fuente’s cross-disciplinary laboratory is using Codex and ChatGPT in an early-stage search for potential antimicrobial molecules within the genomes of living and extinct organisms, OpenAI reported on September 10. The tools help researchers navigate large genomic and protein datasets in which biologically active sequences may be rare.
Candidate selection remains the job of the laboratory’s own deep-learning models. According to OpenAI, those specialized systems identify patterns in biological sequences and prioritize molecules for experimental testing. OpenAI says the computational approach can compress the initial search from years to hours, although its article provides no independent benchmark or detailed evaluation methodology.
Codex and ChatGPT supplement rather than replace the purpose-built biological models. Laboratory members use them to develop hypotheses, write and refine code, download and preprocess datasets, analyze results and compare methods. ChatGPT also helps researchers clarify unfamiliar terminology and connect ideas across biology, chemistry, computer science and engineering.
OpenAI describes the tools as a bridge between disciplines: biologists can get assistance building software, while programmers can more readily engage with biological questions. Researchers may also work with ChatGPT in their native languages, and de la Fuente uses it as a sounding board when shaping hypotheses. This is an account of an ongoing research workflow, not the announcement of a medicine or clinical result.
A model prediction does not establish that a molecule is an effective treatment. Researchers must experimentally confirm antimicrobial activity, determine effective amounts and assess effects on human cells. Promising candidates may then need optimization for safety, stability and effectiveness, alongside evaluation of toxicity, resistance development, behavior in the body and manufacturability.
Even a successful laboratory candidate would still face clinical trials and regulatory review before reaching patients, and OpenAI says AI outputs should be checked for accuracy. The practical significance of the workflow is therefore its division of labor: specialized models rank biological sequences, while general-purpose assistants support programming, data handling and knowledge exchange. That may reduce friction during discovery, but it does not remove the required experimental and medical validation.
Sources
Originally published by AI Feed: OpenAI Details Codex and ChatGPT’s Role in Antimicrobial Research.
Top comments (0)