Dear Ezhil Arasan

First criteria is to select the one which has structural information for your domain of interest. If both the ID's are for same domain, then look for

1. Do the structure is in native form or in ligand bound form

2. The length, for which the structure is solved

3. Do the structure is for a normal or mutated protein

4. resolution and R-factor

5. Closely related species

Depends what you want to do with the structure (just observe or use for, say, molecular replacement). For the latter it's good to look at space groups and how many chains you have in the asymmetric unit. Otherwise, of course, the structure with the better resolution and lower R-factors (as well as other statistics) is better in general - of course you may want to check if loops etc. have been omitted from the other structure or if there are ligands (of biological interest... or not) bound and such things. Two structures of the same protein might have fundamental differences in conformation as proteins can crystallize in different forms in different (or sometimes even in the same) conditions.

Dear Ezhil Arasan

First criteria is to select the one which has structural information for your domain of interest. If both the ID's are for same domain, then look for

1. Do the structure is in native form or in ligand bound form

2. The length, for which the structure is solved

3. Do the structure is for a normal or mutated protein

4. resolution and R-factor

5. Closely related species

In my opinion, there is no "perfect PDB structure", but just the one that better suits your needs. What are you planning to do with the PBD once you have it? What chemical or biophysical question are you trying to answer? Keeping this questions as guidlines will help you select a meaningful PDB. Different entries often correspond to different ligands bound, often mimicking the Michaelis complex or the transition state. Others are trapped in particular conformations by mutations or ligand binding. Also, you may want to look at a particular organism. You need to do some literature research first to find out which one(s) of these PDB are relevant for your research. Then, after you narrowed that down, you can look at other details as number of residues, resolution, and so forth, as subbested above.

Hi

all the discussions, mentioned above are good and useful. In addition those, after taking the intial structure, do simple minimization on the structure using sd or abnr.

Since there is information in each structure, I suggest you consider the superimposition of all structures with a reasonably good resolution and similarity to your protein sequence. Where do they all agree? Where are they most variable? This many tell you more than a single sequence similar structure.