The use of computers with chemical structural information: ICI CROSSBOW system

Abstract

Scientists ask several different types of question involving chemical structures with other property data. ICI has considered these question types in developing the CROSSBOW system for chemical structure recording and search. The system relies on the use of the Wiswesser Line‐Formula Notation to represent structures as compact, logical, unique and unambiguous codes suitable for computer processing. The notations can be converted into detailed explicit descriptions of molecular structure known as connection tables. These allow the most exact atom‐by‐atom search technique for a specified atomic network anywhere in the molecule. To facilitate more general searches, and to provide a screen to reduce input to the exhaustive atom‐by‐atom procedure, two other chemical search techniques are implemented. The first is a fast search for the occurrence of certain chemical fragments, using a fragment set generated from the notations: the second is a search of the notations themselves for symbol strings relating to the required substructure. Output from the search is in the form of structural diagrams generated from the connection tables. Data from associated property files can be displayed along with the structures, or searched as an additional parameter. The CROSSBOW system operates not only on ICI internal files, but also on commercially‐available data bases. These include the Aldrich chemical catalogue, the Index Chemicus Registry System current awareness tapes, and Hansch structure/activity files.

Prediction of biodegradability from chemical structure: Modeling of ready biodegradation test data

“…The substances were characterized by the structural fragments present in their molecules. These fragments were taken from a dictionary of 149 predefined structural fragments [11]. The dictionary was originally developed for the retrieval of chemicals from large databases, and it was designed to achieve a chemical characterization as broad as possible.…”

Review of Estimation Models for Biodegradation

“…The model was based on MITI-I test data for 894 substances of varying chemical structure (388 readily, 506 non-readily). The chemicals were characterised by a set of 127 previously defined structural fragments [42] with the original binary data for biodegradability, the continuous scale was divided into two areas, > 0.55 and < 0.45, corresponding to readily biodegradable and non-readily biodegradable chemicals, respectively. Estimated values between 0.45 and 0.55 were considered as borderline cases.…”

Detailed Review Paper on Biodegradability Testing

“…They considered three types of parameter at the outset: molecular weight; the octanol-water distribution coefficient (K ow ); and Wiswesser line-formula notation (WLN) based on sub-structural keys. They could not obtain accurate K ow values for a sufficient number of the chemicals, so they used only molecular weight as a crude index of bulkiness and WLN-based keys taken from the CROSSBOW program (Eakin, et al, 1974) for describing the chemicals.…”