Models, assumptions, and evidence.

Use Omnis Helix for pre-clinical, in silico software evaluation. Review each model declaration and its evidence before interpreting a score.

Guide scoring

Scores depend on the selected backend, model version, inputs, and assumptions. During a pilot, we record the scorer and compare its output on an agreed in silico evaluation set.

A ranking score is not a measured editing outcome.

Off-target assessment

Site-level assessments depend on the reference, candidate sites, and scoring backend supplied to the run. Review those inputs and limitations alongside the exported evidence.

Coverage and backend availability must be checked for the evaluated build.

Prime and base editing review

Synthetic candidate comparisons demonstrate constraints, declared assumptions, review decisions, and audit export. Their acceptance checks establish software behavior within that demonstration.

Synthetic examples do not establish predictive accuracy on external data.

Verification and provenance

Replayable bundles connect inputs, model declarations, decisions, and verification results. A pilot pins the software version and defines acceptance criteria before evaluating customer-provided inputs.

Artifact verification checks integrity and declared contracts; it does not validate biological performance.