| Summary: | The sequencing of the human genome provides the parts list for understanding cellular processes. However, as 70% of eukaryotic genes work through multi-protein systems, it is only through detailed study of the interactions of these components, that a more complete, systems-level understanding can be gained. This thesis is centred on the establishment of PICCOLO - a comprehensive database of structurally characterized protein interactions. In generating the resource, issues of interface definition, quaternary structure, data redundancy, structural environment and interaction type are addressed. The resource enables a variety of analyses to be performed concerning interface properties including residue propensity, hydropathy, polarity, interface size, sequence entropy and residue contact preference. PICCOLO has been applied to probing the patterns of substitutions that are accepted in protein interfaces across evolution, and whether these patterns are distinguishable from those seen in other structural environments. The derivation of a high-quality set of multiple structural alignments in the form of the database TOCCATA, a prerequisite for such analysis, is described, as well as procedures to derive environment-specific substitution tables. The Blundell group has contributed a series of methods to predict the likely effect of non-synonymous Single Nucleotide Polymorphisms (nsSNPs) on protein stability, function and interactions in order to triage the large volumes of data created from high-throughput genetic screening studies, enabling prioritization of those nsSNPs most likely to be phenotypically detrimental. PICCOLO's contribution to these predictions is described. Historically there has been little focus on protein-protein interactions as drug targets for small-molecule therapeutics. However, alanine-scanning mutagenesis studies have revealed that only a subset of residues contribute the greater part of free energy to binding - so-called 'hot-spots'. Molecular characterization of hot-spots performed using PICCOLO, probes the molecular basis underlying this important phenomenon leading to the possibility of predictive methods to identify hot-spots 'in silico'.
|
|---|
