PROLOG REPRESENTATION OF MOLECULAR-STRUCTURES AND PATTERN-RECOGNITION

Research regarding representation of chemical structures in computers has been performed since the 1970s, with two central goals: information manipulation in chemical data bases and computer assisted molecular synthesis. This work is involved with the latter subject. We show the need for a general flexible and efficient form of representation based on weighted graphs. Some algorithms for structural pattern recognition, such as functional groups, reaction sites, rings, isomorphism and topological symmetry have been developed. These algorithms were implemented in Prolog, a declarative language, used because of its facilities in symbolic computations and its natural back-tracking capability. With the use of this representation we have created a working environment that is simple and effective for the automatic manipulation of chemical structures. This framework also allows easy transformations between different representations, such as connectivity matrices, line notation or topological codes based on path counts, among others, allowing exploration of the particular advantages of each.