Its is most likely that the missing residue could not be observed due to mobility, ie that region is not well resolved, If the region is not so well resolved, this will be evidenced in higher B-factors of the residues surrounding the missing one. While you could add it back in using coot (use the 'add residue'icon, just below the mutate (radiation symbol) icon on the right hand menu, just over half way down), if it is unobserved due to mobility, making a model in which it is frozen in one place is likely to poorly represent the reality.

What is the PDB code of your structure?

Thanks Mathew, PDB ID: 2X8B.

Prior to docking, when I remove all hetro atoms and non-standard aminoscides, then Ser203 also select in this category and finally It removes from structure during pre-docking procedure. I'm wondering why this happens? Although structure has good resolution and R-free factor. See the image below in attachment. 

So If I add that residue, can It be further refine? or its side chain will be perfectly ordered?

Ser 203 is present in the structure, but it is covalently modified by the inhibitor to form O-[N,N-dimethylphosphoramidate]-L-serine (called SEN not SER). When you remove non-standard amino acids, you are deleting it. To keep SER 203, you will need to edit the file to call the residue SER and delete the atoms of the inhibitor

I wouldn't describe 2.95Å as 'good resolution'. Is there any reason you are using 2x8b when there are other structures of human acetylcholinesterase with much higher resolution and better geometry?

All Human acetylcholinesterases as at the following link, sorted by quality

http://www.ebi.ac.uk/pdbe/entry/search/index?ec_hierarchy_name:Acetylcholinesterase&organism_scientific_name:%22Homo%20sapiens%22&sort=overall_quality+desc. 

Even with bound  fasciculin, blocking access to the active site (which 2x8b also has) 4ey8 is higher resolution, slightly better geometry, and not covalently inhibited.

That is great. Thanks for the suggestion. Is there any absolute criteria of selecting protein structures? Lower R-factor and resolution would be enough or other factors should be under consideration. i.e. 

1. to superimpose multiple structures of same protein and select the best representative from them as multiple structures of the same protein might have fundamental differences in conformation as proteins can crystallize in different forms in different conditions or may have the same condition. 

2. By removing the bound ligand from co-crystalline structure and redock it again to check the differences (in RMSD) between the experimental/initial pose and If redocking produce same result then that PDB would be selected for molecular docking protocols. 

What are the key steps of selecting the best out of it?? 

• Thanks. I appreciate your expert suggestions.