Genomarker Diligence
Independent review of computational claims.
For the people who bear the risk of someone else’s computation: investors weighing a thesis that rests on a biomarker result, BD teams evaluating in-licensing data, translational groups inheriting external analyses. Genomarker Diligence answers one question — fast, independently, and in writing: does this result survive scrutiny, and can it be replayed?
How an engagement runs
Claim inventory
Every computational claim in the target material is identified and mapped to the analyses, data, and assurance profiles that could support it.
Assurance evaluation
Each claim’s analysis is reconstructed as a canonical spec and evaluated against versioned rules — methodology, confounding, leakage, plan alignment.
Independent replay
Every examined analysis is re-executed in isolation from the original environment, and the outcome is classified and signed.
Signed verdict report
A claim-by-claim verdict backed by evidence packages, structured findings, and replay attestations — every deliverable verifies offline.
Why independent
The producer of a result — human or AI platform — cannot credibly certify it. Diligence verdicts are issued against named, versioned assurance profiles under a published independence policy: the statement says exactly what was verified, and never more.
What you receive
- Verdict reportClaim-by-claim, signed
- Assurance statementsOne per examined claim
- Structured findingsMachine-readable, rule-versioned
- Evidence packages.gmk · verifies offline, indefinitely
- Replay attestationsIndependent re-execution, classified
SCOPED PER CASE · TYPICAL TURNAROUND: WEEKS, NOT MONTHS
The evidence, not the reviewer, carries the claim.
Every deliverable verifies offline with the open verifier — you do not have to trust our report, or us.