STEREOCHEMICAL AND CONFORMATIONAL CLASSIFICATION OF THE HEXOPYRANOSE SUGARS USING NUMERICAL CLUSTERING METHODS

The methods of single-linkage and complete-linkage cluster analysis have been used to generate, in a rapid and semi-automatic manner, a variety of configurational and conformational classifications of 249 hexopyranose fragments retrieved from the Cambridge Structural Database. Conformational aspects of the fragment were described using standard torsion angles, while configurations adopted at the ring carbons were described by five projected valence angles. The classification experiments show that (a) the pyran ring adopts a C-4(1)-chair conformation in all but ten of the fragments, where it is found that the conformation adopted is that of a C-1(4) chair, (b) only 14 (of a possible 32) hexopyranose stereoisomers are represented in the crystallographic data, and (c) the C6-O6 side chain adopts only +gauche, -gauche and trans conformations. Clustering methods were also used to obtain representative orthogonal coordinates that permit the building of approximate models of hexopyranose units for which crystal structures are not yet available. The analyses indicate that the ten crystal structures having a C-1(4) pyran-ring conformation may be reporting coordinates which describe the wrong enantiomorph. This automated data analysis indicates desirable extensions of the cluster-analysis methodology and also suggests improvements to the evaluation, search and display facilities of the Cambridge Structural Database System.