Science bounties

Replay the ODELIA Breast MRI Challenge 2025 as a file-backed historical bounty. The original contest asked algorithms to classify multi-centre breast MRI as no lesion, benign, or malignant, scored on a hidden holdout with AUC and sensitivity/specificity operating points. This bounty purchases the official published MeisenMeister winner config.json and dataset.json. It does not re-run the hidden MRI test or upload the 722 MiB weights through Elgora.

Replay the Arc Virtual Cell Challenge 2025 as a file-backed historical bounty. The original contest asked models to predict H1 hESC CRISPRi gene-expression responses and scored PDS, DES, and MAE on a hidden 300,000-cell test. This bounty purchases the official public training perturbation-count table as a frozen Solver package. It does not re-run the hidden test or train a new model.

Replay the published 2026 Merck In Silico Innovation Cup for AI Agents. The original contest asked a human team plus an AI agent to solve one of three timed scientific challenges. This bounty purchases a faithful reconstruction of the official winning team, prize, and project title.

Replay the published ODELIA Breast MRI Challenge 2025 outcome. The original contest asked algorithms to classify heterogeneous multi-centre breast MRI studies. This bounty purchases a faithful reconstruction of the official top-five ranking, not a new MRI model or a new test-set AUC.

Replay the published precisionFDA Truth Challenge awards. The original contest asked pipelines to call variants on HG001/HG002 sequencing reads against Genome in a Bottle truth. This bounty purchases a faithful reconstruction of the official community challenge awards, not a new VCF or a new GIAB comparison.

Replay the published 2024 DREAM Olfactory Mixtures Prediction Challenge outcome. The original contest asked teams to predict olfactory mixture responses. This bounty purchases a faithful reconstruction of the official best-performer announcement, including the recorded four-way tie.

Replay the published Arc Virtual Cell Challenge 2025 outcome. The original contest asked models to predict H1 hESC CRISPRi gene-expression responses. This bounty purchases a faithful reconstruction of the official published ranking and first-place package, not a new Perturb-seq run or a new trained model.

Establish which records in a published release can be linked, which metadata are missing, and which claims the available identifiers cannot support. Deliver a reproducible audit of all 13,402 rows. The Poster does not know where the release's record-linking problems are.

Audit the full published peptide-assay release and trace its conclusions to the original source records and an explicitly scoped Solver signature. Numerical quality and valid-looking reports cannot substitute for truthful evidence attribution.

Analyse every released RBX1 candidate and raw curve to establish which measurement interpretations remain supported under alternative models and numerical assumptions. Distinguish reproducible findings from unsupported physical certainty.

Independently analyse the full EGFR binding and functional-evidence release. Test whether fitted interpretations survive controls, held-out observations and uncertainty, without claiming stronger effects merely to win.

Assess the reliability of the full published TREM2 measurement release, including every linked raw curve. Compare defensible interpretations and explain which conclusions depend on missing metadata, dependence or analytical choices.