Publications
bioRxiv preprint — not yet public

Mpro-DockExec: an executability benchmark for SARS-CoV-2 docking claims

NewScience ResearchbioRxiv · 2026

A source-verified, agent-assisted audit of SARS-CoV-2 main-protease (Mpro) molecular docking — the first protocol of the reproduce engine, run by hand.

We gathered 236 open-access docking papers — around 12,466 individual ligand results — and audited every one against a sixteen-field reporting standard built for the job. The picture was bleak: only 8.1% of papers reported enough to run their docking from the methods alone, nearly half were blocked outright by a single missing parameter, and not one of the 236 was fully reproducible.

236
Confirmed Mpro docking papers audited — roughly 12,466 ligand-level claims.
8.1%
Of papers are directly re-executable. 47.9% are blocked by a single missing parameter.
0.36
kcal/mol median agreement on re-docked claims, with 81% within 2 kcal/mol.
κ 0.926
Agreement between two independent reviewers — versus κ 0.075 for an AI-only holistic pass.

Executability class predicts reproducibility: cleanly-reported papers re-docked within a median 0.36 kcal/mol of the published number, while execution-blocked papers couldn’t be reconstructed at all. And the grading itself only holds up when it’s rule-based — a holistic AI-only pass agreed with expert reviewers barely better than chance, while a deterministic classifier reading the same pinned parameters matched two independent experts almost exactly.

Release — Mpro-DockExec: an open, corpus-scale executability benchmark, scoring engine, re-execution scripts, and provenance package.

This preprint isn’t public yet. It’s in final review ahead of a bioRxiv submission. Request access below and we’ll send the full paper directly, before general release.

Want the full preprint?

We’ll send it to you directly, ahead of the public bioRxiv release.