Dear Ben,
thank you very much for your quick reply, I will try your suggestions.

Regards
Micheal

> 1. Use complete_pdb() to make a complete structure for the second PDB (just with "wrong" sidechains).
>
> 2. Make an alignment using the two PDB structures as templates (presumably this a simple 1:1:1 100% identical alignment). Use the output structure from step 1 as the initial model.
> 3. Override automodel's select_atoms() to select only the missing sidechains, and then build ?> comparative models as per usual. Those sidechains should inherit the dihedral preferences from the first PDB.

Dear Lalith,
As I said, I'm really new to modeller but I will help you with this when I figure out how to perform Ben's suggestions.

Regards
Micheal

> Dear Michael.
> I am also working and facing with the same kind of issue through Modeller. I mean I have to copy residues selectively from two different templates for my homology model and have to follow exactly the same steps as Ben suggested to you in the last reply. Here, I am unable to use the initial model for my second strep model generation. I am facing the usual error (saying 'No atoms are specified from input file)  as reported earlier by others. Please let me know if you could successfully execute this step. 
> Thank you in Advance
>
> Regards,
> Lalith