How Xenomine Works

For any protein target with a known structure and characterized binding site, the number of candidate peptide or protein sequences that could interact with it is astronomically large — far beyond what any single laboratory can explore computationally, let alone experimentally.

Current drug discovery concentrates compute in centralized clusters owned by large institutions. Most targets with real unmet need — antibiotic resistance enzymes, agricultural proteins, rare disease targets — sit unexplored not because the science is intractable, but because no single entity has the incentive or resources to run the search at scale.

Xenomine coordinates the compute that already exists — idle GPUs in universities, hospitals, and individual workstations — and directs it at hard biological targets under a shared incentive model.

Miners choose which programs to contribute to. A program is a specific biological target with a crystal structure, a characterized binding site, and a standardized assay for validation.

Programs that matter to communities attract more compute. A lab working on antibiotic resistance can direct their idle GPU at NDM-1. A group interested in crop disease can mine an agricultural enzyme program. The network naturally prioritizes targets that are important to the people running it.

Each miner runs standardized compute jobs — structure prediction, sequence scoring, molecular dynamics — and returns results to the network. Results from all miners pool into a shared sequence database. The more miners, the more of sequence space gets explored.

Each program runs a two-phase compute pipeline. Generate jobs explore new sequence space; evaluate jobs score the most promising candidates in depth.

RFdiffusiongenerates novel backbone structures targeting the program's binding site hotspot residues. LigandMPNN designs amino acid sequences compatible with each backbone. ESM-2 produces sequence embeddings used for deduplication and diversity selection across rounds.

Candidate sequences then undergo evaluation: ColabFold predicts the folded structure of each peptide bound to the target. Molecular dynamics (10 ns implicit-solvent simulation) tests whether the complex holds together under thermal fluctuation.

The composite scoring filter requires pLDDT ≥ 80 (structural confidence), RMSD ≤ 2Å from the designed backbone (fold fidelity), and MD stability above threshold. Only sequences passing all three criteria are submitted to the shortlist for wet lab consideration.

Computational filtering narrows the candidate pool. Wet lab validation determines whether a sequence is real.

The standard assay for enzyme inhibitor programs is a nitrocefin IC₅₀ assay — a colorimetric measurement of whether the peptide reduces enzyme activity at increasing concentrations. A confirmed hit must show dose-dependent inhibition below a program-specific threshold.

For antimicrobial programs, MIC restoration assays measure whether the peptide restores antibiotic sensitivity in resistant bacterial strains. Both assays are run at two independent contract research organizations to eliminate lab-specific artefacts.

Wet lab confirmation — not computational scoring — is the trigger for all downstream legal and commercial machinery. A sequence has to pass the assay before a patent provisional is filed or a program LLC is formed.

Every program has its own token. When you complete a validated compute job for a program, you earn 1 token for that program. Flat rate — no bonus for better hardware, no early-miner premium, no quality multiplier.

Tokens are issued continuously across all rounds, including after a licensing market opens. Supply is uncapped — it grows as long as miners contribute. Early miners are not specially privileged by issuance rate, but they do accumulate more tokens over time by participating longer.

What tokens are worth is entirely contingent on the program producing a licensable asset. If a sequence passes wet lab validation and a licensing deal closes, 80% of all licensing revenue from that program flows to token holders, distributed proportionally to tokens held at the time of each licensing event. 20% goes to Xenomine Inc. as a platform fee, capped in the corporate charter.

If the program is abandoned or never produces a hit, token holders receive nothing financially. The program's sequence data and embeddings are published open-access, and miners are credited as scientific contributors. Open-access publication on abandonment is a deliberate design choice, not a consolation prize.

There is no secondary market in Phase 1. Tokens are ledger entries — not transferable, not tradeable, not redeemable for anything until on-chain deployment is triggered by a confirmed hit.

Token holders vote on program-level commercial decisions. Three decisions are in scope:

• Accept or reject a licensing offer — simple majority. Xenomine presents the terms; token holders decide.
• Pause or resume mining — simple majority.
• Abandon the program and publish open-access — supermajority (>66%). High bar deliberately — this is irreversible.

Quorum is 30% of outstanding tokens. If quorum is not met, Xenomine casts a deciding vote. This is disclosed upfront and exists to prevent governance paralysis on early programs with few token holders.

Token holders in one program have zero governance rights in any other program. NDM-1 tokens vote on NDM-1 outcomes only.

Voting is off-chain via Snapshot initially — no gas cost, wallet-signed. The transition to on-chain binding governance happens per-program when three conditions are met: enough token holders to make on-chain quorum meaningful, the token contract deployed on Base L2 (triggered at first confirmed hit), and legal clearance that on-chain vote execution does not create securities exposure. There is no fixed timeline for this transition.

Xenomine Inc. is the platform entity — a Delaware C-Corp. It owns the infrastructure, the brand, and the 20% fee stream.

Each program that produces a confirmed hit gets its own LLC — a Delaware single-member LLC wholly owned by Xenomine Inc. at formation. The patent provisional for that hit is assigned to the program LLC, not to Xenomine Inc. directly. This separation isolates IP liability per program, makes licensing negotiations cleaner, and creates a clear asset boundary for token holder claims.

LLC formation and patent filing are triggered by wet lab confirmation — not computational scoring. A sequence has to pass the assay before any legal machinery moves. This keeps costs down and ensures the IP being protected is real.

• You are contributing compute in exchange for a contingent stake in a scientific outcome. This is not yield farming. There is no guaranteed return.
• Your tokens are program-specific. If you want diversified exposure, mine multiple programs.
• The science is real, the legal structure is real, the wet lab validation is real. But most drug discovery programs fail. Xenomine's job is to make the search more efficient — not to guarantee hits.
• Open-access publication on abandonment means your work contributes to science regardless of commercial outcome.
• Tokens have no price and no liquid market in Phase 1. Do not contribute compute you cannot afford to treat as speculative.