Difference between revisions of "Homology modeling"
Jump to navigation
Jump to search
Line 17: | Line 17: | ||
==External links== | ==External links== | ||
*An introduction to homology modeling. [http://www.cmbi.ru.nl/~hvensela/EGFR-verslag/homology.html] | *An introduction to homology modeling. [http://www.cmbi.ru.nl/~hvensela/EGFR-verslag/homology.html] | ||
− | *The homology course by the CMBI. [http:// | + | *The homology course by the CMBI. [http://swift.cmbi.ru.nl/teach/HOMMOD/index.html] |
Latest revision as of 09:57, 30 October 2012
Prediction of a protein's structure based on the structure of another homologous protein
Homology modeling can be defined as the step-wise conversion of the structure of one protein into the structure of another protein. The protein to start with normally is homologous to the protein that is to be modelled. The structure to start with is called the 'template' and the structure to be predicted/constructed the 'model'. Homology modeling is done in 8 steps:
- Template recognition and initial alignment: First a correct template (a homologous structure) has to be found. This is typically done by performing a blast-search against the PDB. The best hit is taken as template and a first alignment can be made.
- Alignment correction: The first alignment has to be corrected. A multiple sequence alignment can be very helpfull to decide where the gaps and instertions should be placed.
- Backbone generation: The backbone of the model is generated by copying the backbone coordinates of the template to the model. When a residue is conserved the sidechain can also be copied to the model.
- Loop modeling: Sometimes the model contains more residues than the template so that some ab initio loop modelling is needed. Several approaches exist to create loops. Some of them use libraries of loops with the same residues, created out of all structures in the PDB. Other approaches try to build completely new loops.
- Side chain modeling: When the complete backbone of the model is generated it is time to add the sidechains of the residues. This can also be done by using libraries with preferred sidechain-conformations.
- Model optimization: During the generation of the model many errors are introduced. For example caused by small amino acids that are changed into large residues that bump into other sidechains. In a Molecular dynamics simulation the model is allowed to move and the big errors will be reduced.
- Model validation: Now the [[model] is ready it must be scanned for mistakes. This can be done by hand, or by sending it to several servers, for example the WHATIF server.
- Iteration: It is possible to mistakes force you to go back and try again. Sometimes you have to consider another template and start all over again, sometimes you only have to perform a better molecular dynamics simulation.
The result of the complete process is a model of a so far unknown structure that can be used for mutation analysis, experimental design, etc.