If you have not done so already, you might  find the following paper useful:

Forrest, L.R. et al. (2006). On the accuracy of homology modeling and sequence alignment methods applied to membrane proteins. Biophys. J. 91: 508-517.

What is the quality of model? you can try other homology modeling software like MODELLER or do model refinement by using molecular dynamic simulation or energy minimization, example of software: GROMACS 4.6.3

you can try to find this paper for more information about homology modeling:

Dorn, M., e Silva, M. B., Buriol, L. S., & Lamb, L. C. (2014). Three-dimensional protein structure prediction: methods and computational strategies. Computational biology and chemistry, 53, 251-276.

40% is a descent identity level. Most of the times straightforward single-template modeling works fine. What do you mean by "results are not quite as expected"? If you expect certain residues to be aligned while they don't in your model, you can force their alignment (most homology modeling tools allow that).

If you are looking at homology between two protein sequences with low similarity,try using CD_HIT tool with options 'c 0.4' or 'n 2' as thresholds. Look at user manual for CD-HIT usage.

Good luck.

If the homology modeling results of protein and templates are less than 50% then Ab initio modeling is done.. so try for ab initio modeling may b ab initio modling give u desired results..

I agree with Dmitri Kireev's answer.

Thanks Rudy for your valuable answer.We have decide that instead of modelling the whole protein, we will do so for the peptide(34aa) of our interest from the protein and link the free ends using a potential linker so that they maintain their conformation.

The peptide is showing an identity of `70% with the template and saves time during simulation rather than using the entire protein.Is this a good approach,are there any pitfalls,what the best and acceptable ways of assessing model quality....looking forward to all your answers...thanks everyone

Hello Vineeth,

The generated model can be assessed by the two ways i.e Geometry  and Energy Prospective. There are various ways to go and check the distributions and  clashes of residues (whatcheck, whatif etc..) and energy as well by the dynamics simulations. So, my suggestion is to check your geometry of residues distribution and energy. If above all are satisfied, then proceed for your study or analysis.

Good luck.

the best method is EasyModeller. it is easy to use and pretty accurate