Recent work on GPCR homology modelling found that the modelling of GPCR is different from these cytoplasm protein owing to the low sequence identity with its homology patenters.

Model building by Modeller usually lead to satisfied models based on the protocol provided from Modeller developer. However, when the case come to low sequence identity, the align2D step in Modeller is not so satisfying and the models are not reliable. It seems that this mainly arises from the bad sequence alignment step, and it is logical to improve the result by refining the sequence alignment. Alignment by third-party tools such as: clusterW would be extremely for such kind of purpose.

Of course, other factors such as: secondary structure of 7-TM regions, N-50 residues, SS bond and conserved residues like W, all of which are of great importance for a good GPCR model building, should be also in considerations.

Moreover, it is no doubt that employing refined Modeller protocol and large population sampling will also benefit a lot for the model generation.