Section 22.4.6. Conclusion

This chapter has summarized the vast range of structural knowledge that can be derived from the small-molecule data contained in the CSD. We have attempted to show that much of this knowledge is directly transferable and applicable to the protein environment. Far from being discrete, structural studies of small molecules and proteins have a natural synergy which, if exploited creatively, will lead to significant advances in both areas. It is therefore unsurprising that some of these CSD studies have been prompted by initial observations made on proteins.

As a result of this activity, it is now very clear that software access to the information stored in the CSD and the PDB must be at two levels: a raw-data level and a derived-knowledge level. The onward development of structural knowledge bases from the underlying data provides for the preservation and storage of the results of data-mining experiments, thus avoiding repetition of standard experiments and providing instant access to complex derivative information. Most importantly, a suitably structured knowledge base can be acted on by software tools that are designed to solve complex problems in structural chemistry (see e.g. Thornton & Gardner, 1989; Allen et al., 1990; Bruno et al., 1997; Jones et al., 1997). The availability of knowledge bases derived from experimental observations is likely to be a crucial factor in the solution of those two analogous, and currently intractable, problems in the small-molecule and protein-structure domains: crystal structure and polymorph prediction on the one hand, and protein folding on the other.