PSI Structural Biology Knowledgebase

PSI | Structural Biology Knowledgebase
Header Icons

Related Theme

Model proteins in your lunch break

PSI-SGKB [doi:10.1038/fa_psisgkb.2009.16]
Technical Highlight - April 2009
Short description: How to build yourself a three-dimensional protein model in under 2 hours.Nature Protocols 4, 1-13 (2009)

Protein structures can provide a useful guide to biochemical function and to interaction surfaces and have helped many a biologist to develop meaningful experiments to test biological function. The lack of an experimentally determined structure does not necessarily mean that no structural information is available. The technique of homology modelling can be used to build three-dimensional models of a protein using the experimentally determined structures of closely related members of the same protein family. Producing accurate models in the absence of such a template is very difficult, and at present homology modelling is the most reliable way predicting a protein's structure.

The four main steps of comparative protein structure modeling: template selection, target–template alignment, model building and model quality evaluation.

Homology modelling consists of four steps: identifying evolutionarily related proteins with known structures; aligning corresponding residues; building the three-dimensional model; and assessing the quality of the structure. If the initial model is not good enough, these steps can be cycled through until a sufficiently accurate model is achieved.

Estimating the quality of the model is vital for establishing how reliable it is. There is a clear correlation between the degree of sequence identity between the target and the model and the accuracy of the final model. As a rough guide, the core carbon backbone of models with 50% sequence identity will have a root mean square deviation of about 1.0 Å.

The free web-based software SWISS-MODEL will help you build your model. SWISS-MODEL was the first publicly available automated modelling server, although many more have been developed since. It integrates each of the four modelling steps into one workflow, so if target and template proteins have high sequence identity, you'll hardly have anything to do. At lower levels of sequence identity, you have the option of intervening manually.

Bordoli et al., part of the team behind SWISS-MODEL, provide a simple step-by-step procedure in Nature Protocols explaining how to use this service. And as a bonus they provide troubleshooting tips and worked-through examples. This will make modelling protein structures that have reasonable sequence identity a straightforward task for molecular and cell biologists seeking a structural guide. See Fig. 1

Maria Hodges


  1. L. Bordoli et al. Protein structure homology modelling using SWISS-MODEL workspace.
    Nature Protocols 4, 1-13 (2009). doi:10.1038/nprot.2008.197

Structural Biology Knowledgebase ISSN: 1758-1338
Funded by a grant from the National Institute of General Medical Sciences of the National Institutes of Health