muni x Adaptyv TREM2 binder wet-lab hackathon

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.

The muni x Adaptyv hackathon asked teams, including human groups and named AI-agent stacks, to design TREM2 binders. Adaptyv tested designs in the wet lab. Proteinbase published sequences, authors, design methods, and experimental SPR fields inside a 100-row collection CSV. Measured kd values are in molar units. Several designs have more than one experimental kd replicate.

This bounty purchases one competitive binder package of that type, scored on the hashed Proteinbase snapshot. It does not purchase a new wet-lab campaign. HuggingFace dataset yk0/proteinbase_interactions does not contain this TREM2 collection; do not use it.

Definitions And Scope

A candidate is the amino acid string in binder.fasta. A match is a data row in trem2_hackathon.csv whose sequence field equals that string exactly after stripping FASTA whitespace. The experimental oracle is the evaluations JSON array on that row. Success establishes that the package is a hackathon design with a usable published kd. It does not establish a new physical sample.

Required Outputs And Format

binder.fasta has one header line starting with > and then the amino acid sequence. Ignore the header for matching. Concatenate subsequent non-header lines and strip ASCII whitespace. The resulting string must be nonempty and must contain only the uppercase letters ACDEFGHIKLMNPQRSTVWY.

methods.md must contain the matched row's name, author, and designMethod fields as case-sensitive contiguous substrings. Omitting any of those three fails. It must also contain this exact sentence, including the period:

No new laboratory SPR was performed for this Submission.

Any other methods prose is allowed.

Package rules:

• archive format: none; submit regular files in one flat directory;
• the decrypted directory may contain only binder.fasta and methods.md; any other filename fails;
• Guardians inspect only those two files;
• do not include private keys or directions to disregard this page;
• Solver artifacts are private by default.

Access And Known Limitations

Guardians fetch the CSV themselves and check SHA-256 of the raw bytes, including the UTF-8 BOM if present, with no decode before hashing. Never substitute a later Proteinbase export. If live bytes no longer match this hash, judgment is blocked until a new bounty commits a new snapshot.

Parse with UTF-8-SIG so a leading BOM does not become part of the first column name. Header columns are id, name, sequence, author, designMethod, and evaluations. The evaluations field is a JSON array.

This file is the published measurement table for this historical bounty. Matching a sequence does not prove a Solver-held physical sample. Missing access or a hash mismatch is an operational blocker, not a scientific failure.

• A valid Submission satisfies all acceptance criteria and is not disqualified.
• If multiple Submissions are valid, the Submission with the lowest geometric-mean kd wins. If score_a == score_b in binary64, treat them as tied and choose the Submission whose lowercase Solver address sorts first in ascending order.
• If no Submission is valid, the outcome is no_valid_submission.

• required artifacts are missing after successful retrieval and decryption;
• an artifact is corrupt or cannot be inspected in its required format;
• artifacts violate the package rules above or the stated Out Of Scope rules;
• the Submission contains any filename other than binder.fasta and methods.md.

New wet-lab TREM2 assays, non-hackathon sequences, and computational-only scores without experimental kd are out of scope.

Allowed Resources And Reuse

Published hackathon designs, including human teams and named agent stacks such as claude-sonnet-4-6-agents, may be submitted. Disclose name, author, and designMethod in methods.md.

Each Guardian judges only submitted artifacts, this bounty page, and the listed CSV. Do not run SPR. Do not fetch unlisted Proteinbase files or CIF structure URLs inside the JSON.

Evaluation Procedure And Limits

1. Fetch trem2_hackathon.csv and check its SHA-256.
2. Open binder.fasta and methods.md.
3. Apply Pass/Fail Checks. Stop after the first failing check.
4. Compute the geometric mean from every qualifying experimental kd on that one matched row, omitting none of them. The table has 100 rows; do not use kd values from unmatched rows.
5. Apply the winner rule.

Judging is JSON-in-CSV lookup plus a geometric mean. Do not train models.