NetGoodIndexSubmit a correction

Benefit Ledger · verified · Medicine & Health

Explainable graph networks identify a new MRSA-active antibiotic class

MIT and Broad researchers used ensembles of graph neural networks over 12 million compounds to find a structural class active against MRSA and VRE in mice, published 20 December 2023 in Nature.

20 Dec 2023Tier 3 MajorMethodology 0.1

Current score

+0.35

10 base · Major (tier 3 of 5, 10 pts)
× 0.4500 attribution · Material acceleration
× 0.7000 evidence · Peer review or independent validation
× 0.4500 realization · Experimentally validated
× 0.5500 durability
Event-level product before credit split: 0.78

New structural class with mouse efficacy is major within antibiotic discovery (tier 3), not an approved drug. Modest AI attribution. Peer review. Preclinical realization.

What happened

Models trained on 39,312 measured compounds predicted activity and cytotoxicity; explainable substructure rationales guided purchase and testing of 283 molecules. Two related compounds reduced MRSA burden about tenfold in skin and thigh infection models and showed low human-cell toxicity. Candidates remain preclinical.

Model attribution

+0.35

Chemprop

Predicted antibacterial activity and cytotoxicity to prioritize a structural class.

Models prioritized chemical space; humans measured training data and ran mouse studies.

Attribution 0.4500 · Credit share 45% · Massachusetts Institute of Technology

Claims

  • Deep-learning-guided selection identified a structural class with MRSA activity in two mouse models.

    outcome · supported

Sources

primary sources

Secondary domains: Chemistry

Revision history

  • 13 Sep 2026 · 0.00 → 0.35

    Initial adjudicated seed score under methodology 0.1.