Status: Public documentation
Permea public materials should make computational delivery evidence easier to inspect without overstating what the evidence proves. This document defines the public claim boundary for Permea Core, benchmark outputs, contribution objects, and community-facing documentation.
Permea Core is currently computational and benchmark-first. It provides an open technical foundation for benchmark surfaces, run manifests, evidence cards, and reproducible benchmark workflows.
Permea supports candidate prioritization before wet-lab work. A Permea benchmark result can help reviewers compare sequence-derived delivery-related signal under a defined task, dataset, split policy, feature set, and metric set.
Permea benchmark evidence is not the same as biological validation. Benchmark metrics summarize behavior on a computational benchmark surface; they do not prove transport, mechanism, safety, therapeutic effect, clinical performance, or generalization beyond the defined task.
Public Permea materials may say that Permea is building:
- an open technical foundation for delivery evidence
- benchmark-first infrastructure for reproducible delivery workflows
- computational evidence surfaces for sequence-first biological delivery
- reproducible benchmark workflows for defined delivery-related tasks
- candidate prioritization before wet-lab work
- documentation standards for dataset cards, evidence cards, run manifests, and candidate reports
Public Permea materials may describe sequence-derived delivery-related signal only when the statement is tied to a defined benchmark, dataset, metric set, and limitation note.
Permea does not claim wet-lab validation unless independent experimental evidence has established that claim.
Permea does not make clinical effectiveness claims.
Permea does not claim universal delivery prediction.
Permea does not claim that delivery is solved.
Permea does not claim state-of-the-art performance unless a specific benchmark comparison supports that statement and the comparison is identified.
Permea does not claim public preprint readiness unless the relevant paper is approved and posted.
Permea does not use computational benchmark results as proof of biological transport, mechanism, safety, therapeutic effect, or clinical performance.
Use conservative wording such as:
- computational evidence
- benchmark surface
- candidate prioritization before wet-lab
- sequence-derived delivery-related signal
- reproducible benchmark workflow
- open technical foundation
Avoid wording such as:
- proven delivery
- validated therapeutic effect
- clinical-grade predictor
- universal delivery engine
- solved biological delivery
Before publishing a public claim, reviewers and contributors should check:
- Is the claim tied to a defined benchmark, dataset, metric set, or source?
- Does the statement preserve computational evidence as computational evidence?
- Does the statement avoid wet-lab, clinical, universal, or solved-delivery claims?
- Does the statement avoid unsupported best-performance language?
- Does the statement separate candidate prioritization from biological validation?
- Are limitations, uncertainty, and review status visible where needed?
- Is public preprint or release readiness mentioned only when actually approved and posted?
If any answer is unclear, narrow the claim before publication.