Dear Modellers,

I have used MODELLER to create 100 11-residue long loop models. Some of models cannot form proper loops, so I used their objective function value as cut-off to select loop models. However further biological structure-function studies of the selected models are not all consistent with the experiment results in vitro. Presumbaly, therefore, some of loop models shouldnot have been selected. Meanwhile I threw the crystal structure of the template into MODELLER to model the loop region, which is the corresponding position in the target model. The loop redefinement of the crystal structure show very similar conformations distribution to the model protein. In the crystal structure, 3 H-bonds stabilise the loop conformation. They are E-R , E-S. However R and S are replaced by P and T in the model sequence.

So my questions is what the proper criteria is for this case? Should the conservation of the H-bonds be checked in the loop models? If it is a way how to check H-bonds in 100 loop models automatically before I have some ideas to write a piece of program myself?

Many thanks for hints.

Binbin