Algorithms for the identification of three-dimensional maximal common substructures

Chemoinformatics at the University of Sheffield 2002–2014

“…Sheffield's interest in ligand-based pharmacophore mapping commenced with studies in the late Eighties on the use of clique detection algorithms for the identification of 3D maximum common substructures, [63] with the Bron-Kerbosch algorithm highlighted by this work being subsequently incorporated in the DISCO program developed by Martin et al at Abbott Laboratories. [64] GASP was developed later in collaboration with the Wellcome Research Laboratories, with the program subsequently being distributed by Certara (previously Tripos Inc.) [65] and DISCO, on a test set derived from X-ray crystal structures of protein-ligand complexes.…”

Graph theoretic methods for the analysis of structural relationships in biological macromolecules

“…Following extensive tests, we focused on the use of the Ullmann (1976) and BronKerbosch (1973) algorithms for subgraph and maximum common subgraph isomorphism applications in chemoinformatics (Brint & Willett, 1987a, 1987bGardiner, Artymiuk & Willett, 1998) and have used these as the basis for our subsequent work in bioinformatics; the reader is referred to the original papers for full algorithmic details.…”

Untitled

“…In their respective papers, Levi suggested using a clique detection procedure due to Grasselli [37], and Barrow and Burstall proposed using an algorithm due to Bron and Kerbosch [38]. Until recently [28], the BronKerbosch algorithm was the clique detection procedure of choice for clique-based MCIS applications [1,24,[39][40][41][42][43][44].…”

“…Most of the algorithms described in this paper have been presented from the perspective of arbitrary or 2D graphs, and much of the published literature on the 3D MCS problem simply adapts these algorithms to the 3D case [1,[41][42][43][44]77]. The Barakat-Dean algorithm is an example of an algorithm that was originally proposed in a 3D formulation, and so is the algorithm described by Crandell and Smith [78], although even this owes much to the algorithm of Varkony et al [66], in that it attempts to "grow" a 3D MCIS iteratively.…”