NetGoodIndexSubmit a correction

Benefit Ledger · verified · Medicine & Health

Deep learning identifies abaucin, a narrow-spectrum Acinetobacter antibiotic

MIT and McMaster researchers trained a neural net on in-vitro growth data and identified abaucin, which killed A. baumannii in culture and a mouse wound model, published 25 May 2023.

25 May 2023Tier 2 NotableMethodology 0.1

Current score

+0.08

3 base · Notable (tier 2 of 5, 3 pts)
× 0.4000 attribution · Material acceleration
× 0.7000 evidence · Peer review or independent validation
× 0.4500 realization · Experimentally validated
× 0.5000 durability
Event-level product before credit split: 0.19

Notable AI-assisted lead (tier 2), not a new approved antibiotic. Modest attribution (model ranked compounds; humans screened and validated). Peer review plus mouse data. Realization is preclinical. Durability uncertain without development.

What happened

After screening about 7,500 molecules, the team trained a message-passing model and searched for new actives. Abaucin (related to RS102895) showed narrow-spectrum activity and a mechanism involving lipoprotein trafficking (LolE). It is a preclinical lead, not a human drug. Human chemists and microbiologists designed assays and follow-up.

Model attribution

+0.08

Chemprop

Predicted A. baumannii growth inhibition to prioritize compounds.

Model prioritized candidates from a library; discovery still depended on wet-lab screening and mechanism work.

Attribution 0.4000 · Credit share 40% · Massachusetts Institute of Technology

Claims

  • A machine-learning screen identified abaucin as a narrow-spectrum A. baumannii active with mouse wound efficacy.

    outcome · supported

Sources

primary sources

Secondary domains: Biology, Chemistry

Revision history

  • 13 Sep 2026 · 0.00 → 0.08

    Initial adjudicated seed score under methodology 0.1.