Atom Mapping

Atom mappings can have a large variety of different use cases. The general goal of atom mapping is to find an assignment between two sets of atoms based on a certain motivation. A very common approach is to find an assignment of atoms, between two molecules, which are considered similar/equal leading to an MCS estimation.

For finding such atom mappings multiple different approaches were described in the literature. One is to use 2D isomorphic graph problem-solving in order to estimate the MCS. Alternatively one can use the 3D geometry information of the atoms in order to find such a mapping, like Kartograf does with its atom mapper. Kartograf assumes that the input coordinates of the molecules are highly overlapping due to prior alignment or modeling approaches before the mapping stage is performed. This allows Kartograf to find very efficiently an atom mapping with a minimal atom displacement.

In the case of classical atom mappings for hybrid topology free energy calculations, a core region of the molecules is required. That region is actually, a sub-space of the actual atom mapping and it ensures, that all mapped atoms are connected via covalent bonds in their original molecule, as this might lead to problems in the sampling during the simulations (communicating mapped regions).

Approach:

Atom Mapping Filters

Additionally, rules can be applied during Kartograf’s mapping algorithm, such that for example hydrogens are only mapped on hydrogens or no ring breakage occurs. Such rules might be necessary in order to ensure sampling of physically relevant configurations or serve other purposes. You can find more information and examples in the mapping filters tutorial.