In development
Build small pieces
that can be examined.
Our long-term direction is end-to-end assurance for computational workflows. Bioinformatics provides the first set of concrete questions.
Current research
Specifications for scientific workflows
We are working through how to express selected bioinformatics computations and their assumptions precisely, with human review of their scientific meaning.
The immediate concern is the connection between analysis choices, implementation details and the claims made about a result.
Planned
Small open-source examples
Publish self-contained examples that connect a specification, its implementation and its evidence. Synthetic data will make them practical to inspect and reuse.
Candidate topics include cache dependencies, contrast definitions and multiple-testing family boundaries. These examples are not released tools yet.
Planned
Reusable contracts and review tools
Explore ways to carry data meaning and computational assumptions between workflow stages. Assess where formal proofs improve on tests and runtime checks.
Interfaces, supported languages and proof systems remain design questions. No general Python or R verification claim is being made.
Longer term
Assurance across a complete workflow
Connect verified components without losing their assumptions at the boundaries. Make the remaining trusted parts explicit.
The broader ambition extends to computational knowledge work beyond bioinformatics. A complete, formally verified scientific pipeline remains a research goal.
Follow the work
Join the mailing list for future updates. Confirmation email delivery is still being configured.
We are also preparing a place to discuss these questions in public. Read the community proposal.