Download 19jackson-searchable.pdf

Contents : Searchable database containing comparisons of ligand binding sites at the molecular level for the discovery of similarities in protein function Nicola Gold and Richard Jackson Structural genomics projects produce large amounts of data some of which are solved structures of hypothetical proteins of unknown function. The aim of this project is to aid the characterisation of these proteins by structure based prediction of protein function. The current project extends previous work which demonstrated a method based on geometric hashing to compare the structures and properties of ligand binding sites and assess the extent of their similarity. In particular the binding site of the phosphate moiety of the large class of nucleotide ligands (ATP/ADP GTP/GDP FAD NAD) was studied and is now being extended to include the entire ligand binding sites of these proteins as well as others. This project uses a geometric hashing technique (described previously) which gives a score of similarity a superposition RMSD and equivalenced atoms for each pair of compared binding sites. These data are stored in a World Wide Web accessible database. The prototype database is searchable with a PDB code and ligand information (such as ligand name number and chain). Submission of these data rapidly returns a ranked list of similar ligand binding sites (above a certain similarity score cut-off) with the most similar at the top (Fig. 1). Fig. 1. Example ranked hits with the GDP binding site from Elongation factor Tu protein (1dg1) as the query. Family relatives from the SCOP database are coloured red while superfamily and fold relatives are coloured yellow and blue respectively. Unclassified proteins in SCOP 1.61 are coloured purple. The list of hits details the ligand name number and chain along with the similarity score and the score normalised for binding site size. The final columns contain SCOP codes for the proteins. 1 Future work will see this method and database extended to include ligand binding sites from other proteins and possibly other functionally useful annotations such as PROSITE patterns (database of protein families and domains) and CATH codes (protein structure classification database). We will also implement a statistical measure of similarity. This database and comparison method can then be used to discover new similarities indicating functional relationships between proteins and may uncover binding site similarities in proteins previously thought to be unrelated. It may also be possible to develop a classification scheme for ligand binding sites useful to researchers attempting to identify possible functional relatives. Publication Brakoulias A. & Jackson RM. (2004) Towards a structural classification of phosphate binding sites in protein-nucleotide complexes: an automated all-against-all structural comparison using geometric matching. Proteins Struct. Funct. and Bioinformatics. In press. Funding We wish to acknowledge the support of the BBSRC. Fig. 2. Superposition of the GDP binding site of Elongation factor Tu protein (1dg1) the GDP binding site of signal sequence recognition protein Ffh (1ng1) and the ADP binding site of phosphoenolpyruvate carboxykinase (1k3c). 1dg1 and 1ng1 are from different SCOP families sharing the same superfamily and fold whereas 1k3c is classified in a different superfamily and fold group. The similarity scores and SCOP (Structural Classification Of Proteins) codes of these hits' are also displayed and each hit is coloured according to its similarity to the query's overall fold and SCOP family thus showing the nature and relative interest of the hits in a visual and easily assimilated layout. In the example shown in Fig. 1 a highly similar binding site has been found which has a different overall fold to the query and is easily identifiable as such because of its colour. The next stage of the program calculates optimal superpositions of the binding site and ligand of interesting hits on the query allowing further examination and visualisation with molecular graphics packages (Fig. 2). A multiple alignment of structurally equivalenced atoms is also provided. 2

INFO HASH : 13b049dee7b43002a6780b124634512425513aaa

Download now

File Size: 159.7 kb