We have developed GPCRModSim, a web server that facilitates the process of modeling GPCRs using Modeller as a back end modeling tool (http://gpcr.usc.es).
In particular, once you upload your sequence or swissprot code, you will get an optimized alignment with all crystallized GPCRs (you can choose also between templates in the active or inactive conformations). The server will recommend you the most suitable template based on TM homologies, although you can further inspect this and change the default selection on the basis of homologies between particular regions.
For a test case, the H1 histamine receptor modeled with excelent accuracy, see the tutorial at http://gpcr.usc.es/tutorial
There are several other features, like the possibility of loop refinement (using loopmodel routines) and the MD simulation of your simulated model in an atomistic model of the membrane (using GROMACS routines).
Hugo
-- Hugo G. de Teran, PhD. "Parga Pondal" Research fellow Fundación Pública Galega de Medicina Xenómica - SERGAS Santiago de Compostela (SPAIN)
I am interested in using Modeller for predicting the structure of
the G protein-coupled receptor I study. This protein does not have any
homologous proteins in the PDB database and there is no experimental data to
help model the protein. Knowing this, can Modeller provide an accurate
structure of my protein to use in molecular dynamics simulations? I
understand that I can choose other crystallized GPCRs to use as templates, but
I am concerned that since Modeller is a homology modeling program that the end
model would simply look like my protein forced onto the template structure.
If anyone can give me some further insight on this manner, I would
greatly appreciate it.