Hi team,
I've developed a lightweight, pure-Python implementation of the Minimum Cycle Basis (MCB) algorithm, capable of processing graphs with over 100,000 edges while maintaining memory usage below 2 GB through sparse matrix optimization.
In complex fused-ring systems (e.g., polycyclic aromatic hydrocarbons), traditional ring perception methods (like those in RDKit) may miss or misrepresent ring bases. MCB provides a mathematically rigorous solution by extracting the minimal set of fundamental cycles.
The implementation is open-source and includes performance benchmarks and usage examples:
🔗 https://github.com/m15154178071-cmyk/graph-algorithms
If this aligns with your roadmap for reacnetgenerator, I’d be happy to discuss integration, provide a PoC, or contribute a module.
Thanks for your great work!