The Quest for Computing Disconnected Maximum Common Substructures under Constraints

26. Januar 2021

Foto: Robert Schmidt

The maximum common substructure (MCS) problem has many applications in science. In cheminformatics, it is used to detect molecules with common building blocks and scaffolds. So far, algorithms exist for the connected and disconnected variant of the MCS. We now develope an algorithm named RIMACS able to control the number and size of connected components precisely. Our new publication shows the use of RIMACS on molecule data as it is typical in drug design.
Extensive evaluations show good runtime behavior outperforming other up-to-date open-source MCS algorithms on the connected MCS problem. The RIMACS algorithm and its corresponding open-source implementation is released on our GitHub account https://github.com/rareylab/RIMACS. The full paper with all the details can be found here.