Hi, 

BLAST search shows 45% identity and E value =7. Not significant.

Try to model using Phyre 2...it may give you best result. 

Also try RaptorX and ITASSER

Thank you i will try that

all depend where are the missing residues. If there are in the loop on the surface of the protein, then you can use any approach. If the missing residues are inside of the protein and correspond to a secondary structure, then you should identify a homologue 3D folding and build the missing part. 

A modelling tool using homology modelling or threading method will give you a model based on structure databases available, ITasser or any other ab-initio based methods will help you simulate your model.

As far as the missing amino acids are concerned, SCOP or CATH will help you to classify the protein and in a way you can add amino acids.

Hi, you can perform a homology modeling of the complete protein and subsequently, refine the cgenerated loop by Prime refinement environment (using the parameters for a loop with 10 or more residues). Moreover the optimized model could be submitted to the  MD simulation for obtaining a reliable structure for further studies.

You can also use Robetta server (http://www.robetta.org/submit.jsp) which I feel is the best for modelling of the proteins for which significant homologues are not available. It builds the parts of the model based on homology, and some parts by ab-initio modelling and threading. I hope it will help you. 

firstly check physiological parametres then go to modelling

In Molecular Operating Environment (MOE) software can overcome the problem of missing residues but that software is expensive. So I would suggest I-Tasser.