Enumerating common molecular substructures

Abstract

Finding and enumerating common molecular substructures is an important task in cheminformatics, where small molecules are often modeled as molecular graphs. We introduce the problem of enumerating all maximal k-common molecular fragments of a pair of molecular graphs. A k-common fragment is a common connected induced subgraph that consists of a common core and a common k-neighborhood. It is thus a generalization of the NP-hard task to enumerate all maximal common connected induced subgraphs (MCCIS) of two graphs, which corresponds to the k = 0 case. We extend the MCCIS enumeration algorithm by Ina Koch and apply algorithm engineering techniques to solve practical instances fast for the general k > 0 case, which is relevant, for example, for automatically generating force field topologies for molecular dynamics (MD) simulations. We find that our methods achieve good performance on a real-world benchmark of all-against-all comparisons of 255 molecules. Our software is available under the LGPL open source license at https://github.com/enitram/mogli.

OFraMP: a fragment-based tool to facilitate the parametrization of large molecules

“…To find all possible matching fragments within a pair of molecular graphs, the problem to be solved is Enumerating all Maximal Common Fragments ( k -MCF–E). It can be shown that the k -MCF–E problem is a generalization of enumerating all maximal Common Connected Induced Subgraphs (CCIS) (MCCIS–E) [ 29 ]. Note, there is a fundamental difference between enumerating all maximal CCIS (MCCIS–E) and the well-known problem of finding the maximum (largest) CCIS.…”

“…For OFraMP the MCCIS–E algorithm of Koch [ 33 ] was adapted to k -MCF–E. We have shown previously that when combined with additional data reduction techniques, this algorithm for solving k -MCF–E is highly efficient even for large molecular graphs [ 29 ], allowing databases containing 100,000s of molecules to be screened within seconds. A detailed description of the algorithm used is provided as Supplementary Information (SI).…”

Automated partial atomic charge assignment for drug-like molecules: a fast knapsack approach

“…In previous work, we have contributed to improving the reliability of this repository by ensuring the consistency and utility of the partial charges assigned to atoms by identifying atoms that could be used to form charge groups , which can be collectively assigned integer formal charges () [10]. We have also developed methods to match molecular substructures, taking into account that the partial charge of an atom is heavily dependent on its neighbours and the nature of its local chemical environment [11]. This made it possible to study the distribution of charges within local molecular environments for all molecules in the ATB ( 260,000 molecules; 9,000,000 atoms and 9,100,000 bonds) and to find, given a query molecule, all possible matching fragments (sub-graphs).…”