"Shape signatures" is a novel technique for generating compact descriptors of molecular shape and of the
shape of receptor sites. This is achieved by using a technique much like ray-tracing to explore the volume
defined by the surface of a molecule or, alternatively, the volume external to a protein in the vicinity of
a receptor. Probability distributions are derived from the ray trace and used as compact descriptors of shape.
Databases of chemical compounds can be augmented with these shape descriptors and used in both ligand- and
receptor-based molecular design.
Biomolecular display technologies enable the high-throughput analysis of protein interactions, however,
functional display of diverse genomic mammalian proteins usually requires native machinery for correct folding
and post-translational modification, machinery that is absent from prokaryotic and in vitro display systems. We
are developing a display technology platform, that creates protein display libraries using native expression systems,
including human cells, where each expressed protein is covalently linked to its corresponding cDNA. The sensitivity of
detection, accomplished by sequencing of the cDNA, is many orders of magnitude better than direct detection of protein
using mass spec or 2D gels. All expressed proteins in a cell can be screened for binding interaction with a compound
in a single experiment using this display technology. We screen small molecule drugs against human protein libraries,
linking chemical structure directly to protein sequence. This approach complements expression profiling and traditional
proteomics methods in the effort to integrate bioinformatics with chemoinformatics.
