Searching SOCRATES

SOCRATES is used widely by Pfizer scientists in a VAX cluster computer environment at Sandwich and is well regarded both for its ease of use and its range of integrated search capabilities, allowing combinations of a variety of chemical and biological search parameters. 

With the graphical output program in place, it was possible to provide our scientists with the first fruits of the early SOCRATES system graphical structure display at a terminal of compounds from the company files (generated fi om the stored WLN s via connection tables). The next step was to provide a structure match capability. As mentioned earlier, within INDABS there already existed a fast chemical novelty search which employed an index of topological codes. This was easily integrated into SOCRATES to accept graphically drawn structures fi om SENTRY instead of WLN s. 

SOCRATES search system (as shown schematically in Figure 12). It is also used within the chemical search subsystem itself to provide offspring searches and NOT logic capabilities (see searches illustrated in Figures 6 13). 

An extension of SOCRATES to handle chemical reactions, called CONTRAST, has also been developed at Sandwich, again with some collaboration from Peter Willett at Sheffield. In SOCRATES the precursors of our company compounds are stored in the compound registry file, and so application of a reaction site detection algorithm to the starting material and product generates a set of modified connection tables and fragment screens, and a set ofreaction bit-screens. The substructure search graphical structure input menu was modified to enable the chemist to identify the atoms involved in the reaction, but otherwise is essentially the same interface as in SOCRATES. 

Other software that has been recently integrated into SOCRATES includes FLATLAND, a package designed to incorporate chemical structures and schemes into word processing systems, and 3D-SEARCH, which extends our searching capability into databases with 3-D structure coordinates (e.g., the Cambridge Crystal Structure... 

This is used within SOCRATES in two ways either as a browsing search, an alternative to a conventional substructure search, on a full file, or as a means of making a large output from a substructure search more intelligible, by ranking the retrieved structures in order of similarity to an ideal target structure. 

The procedure in both cases is the same, and has been described fully elsewhere. The target or ideal structure is input, using the standard SOCRATES graphics front-end, or simply specifying a name, laboratory code number etc. The procedure thereafter is fully automatic, so that this search method is immediately accessible to the least experienced user. 

Several parameters of the procedure can be varied, and all have been exhaustively evaluated and compared the methods adopted in the operational system reflect the best solutions to emerge from these evaluations, with some compromises to improve efficiency. The standard set of 1315 fragment screens, chosen on frequency grounds for use in SOCRATES substructure searching, is used as the structural features, and only presence/absence information, rather than occurrence counts, is used. The Tanimoto... 

In a similar fashion to structure browsing, this is used in two ways within SOCRATES. A complete file of structures may be divided into clusters, systematically reflecting the structural variation within the file, and one compoimd chosen from within each cluster, so as to create a subset of structural representatives , e.g., for use in large-scale screening programmes. Alternatively, a substantial output from a substructure search may be clustered, and one exemplifying structure from each cluster examined. In this way, the structural variation within the output may be readily appreciated. 

The molecules in these data-sets were then characterised by the presence or absence of the screens that are used for substructure searching in SOCRATES, the Pfizer Central Research chemical and biological databank system . In the case of the 2-D structures, there were 1315 atom-centred, bond-centred and atom sequence fragment screens in the case of the 3-D structures, there were 1471 inter-atomic distance screens. ... 

---