|
|
| |
| |
| Modelling Structures of Proteins and their Complexes |
| |
Our aim is to accurately model the 3D structures of proteins and their complexes. To this end, we combine the laws of physics along with experimental observation and statistics to develop computational methods in structural biology. The methods are tested, often in close collaboration with experimental biologists, on particular systems of interest. Our research gives detailed information of cellular processes and hence impacts research on human health, nutrition and biology as a whole.
Predicting protein-protein interactions
The broad aim of our group is to develop and apply computational tools to model the structural biology of inter-molecular interactions in the living cell.
As a first step, protein complex compositions are predicted and modeled by structural similarity to known (template) protein domain-domain interactions [Davis et al 2006, Pieper et al 2005]. The complexes predicted are not restricted to pairs of proteins. If multi-domain templates are present, multi-component interacting protein complexes are predicted. The complexes are assessed using a statistical potential constructed from residue contacts across known protein domain-domain interfaces. Prediction scores are calibrated for reliability. On a benchmark set of 100 interactions, the statistical potential accurately predicted interactions in 97 cases. The method is also capable of distinguishing between alternate modes of binding (see Figure 1). Additional information such as functional annotation and sub-cellular localization can be used to enhance reliability.
|
| |
|
| |
| Figure 1: Camelid VHH domains AMB7 (orange), AMD10 (magenta) and AMD9 (blue) bind to porcine pancreatic a-amylase (PPA, gray surface) through three distinct binding modes (PDB codes 1kxt, 1kxv, and 1kxq, respectively). All three modes were evaluated for each VHH–PPA complex using the interface statistical potential. The statistical potential is sensitive enough to distinguish the native binding modes from the non-native modes. |
| |
| |
We are now developing methods that
a) Model protein interactions without the aid of full length (entire domain coverage) templates.
b) Model all biological complexes, not restricted to protein interactions.
|
| |
References
- Davis FP, Braberg H, Shen MY, Pieper U, Sali A, Madhusudhan MS.
Protein complex compositions predicted by structural similarity.
Nucleic Acids Res. 2006 May 31;34(10):2943-52. Print 2006.
PMID: 16738133 [PubMed - indexed for MEDLINE]
- Pieper U, Eswar N, Davis FP, Braberg H, Madhusudhan MS, Rossi A, Marti-Renom M, Karchin R, Webb BM, Eramian D, Shen MY, Kelly L, Melo F, Sali A.
MODBASE: a database of annotated comparative protein structure models and associated resources.
Nucleic Acids Res. 2006 Jan 1;34(Database issue):D291-5.
PMID: 16381869 [PubMed - indexed for MEDLINE]
|
|
| |
|
