Computer storage and retrieval of generic chemical structures in patents. 7. Parallel simulation of a relaxation algorithm for chemical substructure search

Abstract

A relaxation algorithm for chemical substructure search is simulated for implementation on general-purpose multiprocessors. An improved relaxation algorithm is described and the inherent parallelism detailed. The general-purpose simulation package PASSIM is described, and the methods used to simulate the algorithm are given. A variably sized pool of processors was assumed. The simulation was run on 71 structure/query pairs, and an average maximum speedup of 5.5 over a single processor was found, for approximately 20 processors. A great variation is found for individual structure/query pairs. The overall factor limiting the performance is the serial bottlenecks in the algorithm.An isomerism between polycyclic aromatic hydrocarbons and ring assembly compounds formed by replacing selected rings with acetylenic linkages was noted, but not developed, by Dias. A far more complete isomerism, which is the logical extension of the referenced isomerism, is

Computer storage and retrieval of generic chemical structures in patents. 9. An algorithm to find the extended set of smallest rings in structurally explicit generics

“…(1,2,9,10,11,16), (1,13,7,12,11,6), (14,10,9,2,3,4), and (14,10,11,6,5,4), to leave the bond-used values as shown in Figure 7ii. The class II cut face (1,2,9,10,11,16) has been found to the detriment of the simple face (7,8,9,10,11,12), which now has two adjacent edges with bond-used values of 2 and so cannot be traced by further processing. However, vertex 1 has a ring connectivity of 3, two incident edges, ( (1,13), with a bond-used value less than 3.…”

“…However, for the second alternative, the paths (1,7,8,4) and (1,2,3,4) are of equal length, and both must be designated CHILINK: SHORTEST; there is no CHILINK:MAXIMAL for this expansion. Furthermore, for the third alternative, the classification is reversed; the path containing Rl is now longer than the other path, and so (1,9,10,11,4) is designated CHI-LINK:MAXIMAL, while (1,2,3,4) is the CHILINK: SHORTEST. Overall, the CHILINK:SHORTEST for R2 + R3 will contain a 3-vertex áll-carbon, a 4-vertex all-carbon, and a 4-vertex heteroatom path, while the PARLINK: ROLLING list will contain a 4-vertex heteroatom and a 5vertex all-carbon path.…”

“…In place of the atom-by-atom search, a relaxation algorithm search strategy13 is used and has been further evaluated in a parallel processing implementation. 11 In addition, there is the technique of simplifying structures to the reduced-graph form. 12 This paper reports a ring perception algorithm developed to find the ESSR in structurally explicit parts of generic structures, i.e., all ECTR leaf nodes containing connection tables.…”

Parallel Algorithms for Graph Cycle Extraction Using the Cyclical Conjunction Operator

“…Although the algorithms are not equally efficient on any type of parallel computer. 5 Computer-Assisted structure elucidation (CASE), 6 3D structure prediction, 7 calculation of molecular electronic structure properties, 8 management of chemical structure database, 9,10 searching DNA sequences, 11 and others much, are some of the multiple applications of parallel computing on chemical sciences. 12 The extraction of the present system of rings in a graph is another complex problem that requires a high computational effort.…”

The Sheffield Generic Structures Projecta Retrospective Review

“…This progressed well; the British Library R&D Department funded continuation of the initial doctoral researchers from 1981 until 1983, at which point finances became available from both Chemical Abstracts Service and from Derwent Publications Ltd., with support from the latter continuing for several years. A series of publications documented progress in the project. − …”