Molecular similarity aims to give a quantitative answer to the question of how similar two given molecules are. Such indices are of use in drug design as aids to the creation of molecular mimics and in structure-activity studies or measures of diversity. Similarity is most often computed in terms of molecular shape or electrostatic potential.

The advent of combinatorial techniques and the use of high-throughput synthesis have created a demand for ever faster methods of computation. Numerical calculation has been superceded by analytical evaluation of integrals, but even faster methods are urgently needed. This is especially so if we can ever hope to take thousands of molecules and to calculate the similarity between all pairs.

A promising technique is to use two-dimensional molecular representations and to utilize methodologies perfected in optical character recognition.

Back to Workshop Schedule

1996-1997 Mathematics in High Performance Computing