Hi,
We've been working on K-Lean (https://github.com/Heime-Jorgen/kenosian-lean4) — a Lean 4 library providing machine-verified mathematical contracts for computational biology.
Given Reactome's work on the graph-based biological pathway knowledge graph, a few contracts may be directly relevant:
- KLean.Pathway.ParticipantConservation — proved that reaction participant counts (inputs + catalysts) are conserved under the stoichiometric model: total molecular entity count is non-negative at each reaction step
- KLean.Graph.HierarchyDAGInvariant — machine-verified that the Reactome pathway hierarchy forms a valid DAG: no cycles exist under the hasEvent/hasMember containment relation, making topological sort always defined
Each contract is sorry-free and verified by the Lean kernel — meaning the mathematical properties (bounds, monotonicity, conservation laws) are machine-certified, not just tested.
Happy to hear if formal verification would be useful here, or if there are specific formulas in this codebase worth covering.
— Kenosian (https://kpp.kenosian.com/k-lean.html)
Hi,
We've been working on K-Lean (https://github.com/Heime-Jorgen/kenosian-lean4) — a Lean 4 library providing machine-verified mathematical contracts for computational biology.
Given Reactome's work on the graph-based biological pathway knowledge graph, a few contracts may be directly relevant:
Each contract is sorry-free and verified by the Lean kernel — meaning the mathematical properties (bounds, monotonicity, conservation laws) are machine-certified, not just tested.
Happy to hear if formal verification would be useful here, or if there are specific formulas in this codebase worth covering.
— Kenosian (https://kpp.kenosian.com/k-lean.html)