Marsden, Russell Leonard; (2003) Analysis and prediction of protein domains. Doctoral thesis (Ph.D), UCL (University College London).

Text Analysis_and_prediction_of_pro (1).pdf Download (10MB)

Abstract

Domains are the structural and functional units of protein structure enabling the formation of multi-functional, multi-domain proteins. The identification of domains within a sequence is an important pre-requisite for many protein analysis techniques and can be achieved from protein tertiary structure or detection of homology to known domain sequences. In the absence of either protein structure or sequence homology a method to delineate domain boundaries from sequence alone is required. In this thesis, a number of domain properties have been considered in order to address the possibility of domain prediction from sequence. A survey of domain linker characteristics has been made which shows domain linkers to be flexible, exposed and generally unstructured regions of polypeptide, with a high propensity for proline residues which may have repercussions for linker structural independence and folding rate. The surface area and hydrophobicity of protein structures has also been investigated. There appears to be a positive correlation between sequence length and surface area, although a domain prediction method based solely on this characteristic does not seem likely. There was no obvious separation between the percentage of hydrophobic residues in either single or multi-domain proteins. A domain assignment method based upon the alignment of predicted secondary structure to proteins of known structure has been developed and implemented. The top-hit prediction of continuous domain boundaries achieved a sensitivity of 31% with a selectivity of 32% (± residues). The domain number and corresponding boundaries were correctly predicted for 25% of the multi-domain test set (±20 residues). A further method that assigns domains based on post-processing PSI-BLAST alignments has also been developed. This method achieved a continuous domain boundary prediction sensitivity of 30% with a selectivity of 56% (±20 residues). These two methods have also been combined for prediction of domains by sequence comparison and from sequence alone. The formation of protein oligomers by the exchange of identical units of protein structure between subunits is termed 'domain swapping'. A general analysis of domain swapped proteins, including the properties of the swapped-domain linkers has been carried out. The analysis of domain linker of Chapter 2 also enabled a comparison of their characteristics to those of swapped-domain linkers.

Download activity - last month

Export as XMLCSVJSON

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as XMLCSVJSON

Archive Staff Only

View Item