Kartograf: An Accurate Graph-Based Atom Mapper For Hybrid Topology Relative Free Energy Calculations

Relative binding free energy (RBFE) calculations have shown over the years to be an incredibly
powerful tool in supporting ligand optimization efforts. Despite many successes, the use of
RBFEs can often be limited by problems with applying these approaches, in particular
automating calculation setup procedures.
Atom mapping algorithms are an essential component in setting up hybrid topology RBFE
calculation campaigns, particularly when applied at large scales. Traditional algorithms typically
use a Maximum Common Substructure approach, which has limitations when dealing with
chemical properties related to geometry, and can lead to suboptimal solutions.
To overcome these limitations, we have developed kartograf, a Python package for the planning
of free energy networks which features a geometric-graph based algorithm. The algorithm uses
primarily the 3D coordinates of atoms to find a mapping between two ligands. Usually in free
energy approaches, the ligand conformations are derived from docking or other previous
modeling approaches giving the coordinates a certain importance. By considering the spatial
relationships between atoms related to the ligand poses, our algorithm can bypass the
computationally complex subgraph matching approach of MCS and reduce the problem to a
much simpler sparse graph matching problem. Additionally, this algorithm circumvents typical
mapping problems induced by molecule symmetry and stereoisomerism, making it a more
robust approach for atom mapping from a geometric perspective.
We validate katograf against current gold standard methods, namely Lomap, applying it to the
calculation of relative hybrid free energies for a diverse set of small molecules. Our results
demonstrate that kartograf offers an interesting alternative approach.
The code for kartograf is freely available on Github
(https://github.com/OpenFreeEnergy/kartograf). Kartograf has been developed for the OpenFE
ecosystem, but can also be used as a standalone python package.