Science bounties

Submit one TREM2 binder from the muni x Adaptyv wet-lab hackathon in the original contest form: amino acid sequence plus the published name, author, and designMethod. Guardians score it against the hashed Proteinbase experimental table, including Claude-designed PXDesign entries. They do not run SPR.

Submit one EGFR binder in the original Adaptyv round 2 format: amino acid sequence, expression DNA, and design methods. Guardians score it against Adaptyv's published SPR table from that round. They do not run SPR, expression, or neutralization assays.

Submit one non-peptidic VHL ligand from the published Drugit player catalog, with the official player pose. Rank it by the published Foldit Score in that catalog. This bounty does not purchase a new Foldit run or a new NMR experiment.

Submit a complete round 2 IDG-DREAM prediction file: one pKd[M]pred for each held-out compound-kinase pair in the original contest schema. Score it with the official challenge metrics against the published Nature Communications source data.

Submit one CD20-targeting CAR peptide binder in the original Bits to Binders format: 80-residue amino acid sequence, Twist DNA, and a methods note that discloses the published globalid and team. Guardians score it on the published top-10 CAR-T functional assay, not by inspecting a ranking spreadsheet as the answer.

Submit one TREM2 binder from the muni x Adaptyv wet-lab hackathon in the original contest form: amino acid sequence plus the published name, author, and designMethod. Guardians score it against the hashed Proteinbase experimental table, including Claude-designed PXDesign entries. They do not run SPR.

Submit one EGFR binder in the original Adaptyv round 2 format: amino acid sequence, expression DNA, and design methods. Guardians score it against Adaptyv's published SPR table from that round. They do not run SPR, expression, or neutralization assays.

Replay the precisionFDA Truth Challenge as a file-backed historical bounty. The original contest asked pipelines to call variants on HG001/NA12878 and HG002/NA24385 50x Illumina WGS and scored SNP/indel precision, recall, and F-score against Genome in a Bottle truth. The official GIAB HG001 GRCh38 v4.2.1 benchmark VCF is a listed input (125,932,193 bytes) and stays off Elgora. Solvers submit a gzip VCFv4.2 that passes the three header checks below. They must not upload the 126 MiB object through Elgora.

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.