Off-Target Research Evidence Scoping
Define a bounded off-target evidence evaluation without representing unsupported synthetic-biology runtime capability as active.
Who this is for
Genome-editing research leads, synthetic-biology safety reviewers and technical-diligence teams scoping bounded off-target evidence.
The problem you already have
Off-target risk is multi-factor, but the dedicated runtime is not yet proven
Part 121 discloses CRISPR off-target evaluation using several evidence dimensions — sequence mismatch structure, PAM compatibility, chromatin accessibility and conservation. That gives a concrete technical contract to validate, but it does not change the current truth-lock: this page remains scoping-only until a dedicated runtime path is proven.
Sequence similarity alone can overstate or understate the concern
The Part 121 disclosure distinguishes mismatch position from mismatch count and separately includes PAM and chromatin context. For scoping, that means the evidence plan should preserve which dimensions are available and which are missing rather than collapsing every candidate site into a generic homology score.
A computational off-target score is not a genome-editing safety determination
Even if a future supported path implements the disclosed risk tiers, the output would still require experimental validation and qualified human review. Sufficient evidence for a research screen must never be represented as validated clinical editing safety or treatment authority.
Real-World Operational Scenario: The Guide That Looked Clean Until Chromatin and PAM Context Were Added
Scenario: A candidate guide looks acceptable under a simple mismatch-count review, but several plausible off-target sites have stronger contextual evidence.
A genome-editing research team wants to know whether a deeper off-target evaluation is technically supportable before committing to a larger validation program.
WITHOUT this scoping gate: The team may either over-trust a sequence-only screen or over-build an evaluation before confirming that the required PAM, accessibility, conservation and reference-genome evidence can be supplied consistently.
WITH this scoping gate: The proposed evaluation contract explicitly identifies the required evidence dimensions, acceptance criteria, missing fields and implementation status. If a supported runtime cannot be proven, the result remains SCOPING_ONLY rather than becoming an off-target safety score.
What BivectorAI Provides
Off-Target Evidence Contract
Defines the minimum research evidence needed to evaluate a guide/off-target question, including which sequence and contextual dimensions must be present before an assessment can be considered supportable.
Context-Aware Validation Design
Uses the Part 121 disclosure to scope validation around mismatch structure, PAM compatibility, chromatin accessibility and conservation without claiming those mechanisms are already exposed as a deployed product runtime.
Supported-versus-Unsupported Decision
Separates SUPPORTED, PARTIAL, UNVERIFIED and SCOPING_ONLY outcomes so patent coverage cannot be mistaken for implemented customer capability.
Research-Only Safety Boundary
Keeps clinical editing authority, patient treatment decisions and validated genome-editing safety claims outside scope even if a future private evaluation path is established.
Prove the Gate Before Broader Integration
Validate the evidence contract first; prove the runtime second.
Freeze the guide/off-target input schema, required contextual fields, candidate-site labeling rules, acceptance criteria and experimental reference plan. Then require a reproducible implementation path to execute that same contract. If either the evidence or runtime proof is missing, broader integration remains blocked and the commercial result stays scoping-only.
What you provide
A redacted research question, the available guide/off-target evidence and the required validation boundary. Where already available in your research workflow, identify the reference-genome context, mismatch structure, PAM context, chromatin-accessibility evidence and conservation evidence that a future supported evaluation would need. Clinical editing authority and patient data are not required.
What BivectorAI actually hands you
Bounded evidence contract
The exact proposed input fields, evidence dimensions and question boundary required for a defensible off-target research evaluation.
Acceptance and validation criteria
The conditions that must be met before a result could be treated as a valid research screen, including the experimental reference boundary.
Capability truth decision
A clear SUPPORTED, PARTIAL, UNVERIFIED or SCOPING_ONLY determination based on available implementation evidence.
Private evaluation plan
A bounded next-step plan only if the runtime and evidence path can be proven; otherwise the record explicitly stops at scoping.
What we will never ask you for
Off-target safety questions can sit close to clinical and genome-editing authority, but the current product is scoping only. The engagement is limited to evidence and validation boundaries, so clinical authority and unsupported safety claims remain outside scope.
- Clinical editing authority.
- Patient treatment decisions.
- A validated genome-editing safety claim.
- A representation that the dedicated product runtime is currently proven.
How it actually runs
Start with a bounded evaluation
Fixed scope / quoted per engagement. Inputs, outputs, evidence boundary and acceptance criteria are agreed before work begins.
We do not open with an annual license. If the bounded scoping review does not establish a supported evidence contract or evaluation path for the agreed off-target question, there is no reason to move into a broader commercial commitment.
Off-target variables exist in related synthetic-biology research code, but the corresponding public synthetic product surfaces are currently fail-closed and this truth-lock did not prove a dedicated off-target runtime. This page therefore represents scoping only.