Structure Validation: Softwares & Procedures

[karthik3k]

Structure Validation ...

What are the softwares/procedure used to validate the PDB Structure obtained from various softwares like Swiss Model ??

 

[Monique]

The ultimate validation would be to make a crystal of the compound and do some X-ray diffraction. It is extremely difficult to predict the structure of proteins. Usually the information is taken from molecules, to which it is similar, from which the structure IS known.

So I guess you looked up a molecule in Swissprot and you want to compare it to another database.. I'd first find out how that original PDB structure was modeled..

 

[Monique]

How about the http://deposit.pdb.org/mmcif/VAL/"?

 

[iansmith]

One thing you can do using bioinformatics use different software for tertiary strucutre prediction and id the common structure. Expasy Protein Tools has several link to various tertiary structure predecition tools. If 90-100% of the predict a certain strucutre within a certain region, then prediction is kind of valid and is like to occur.

But as monique said the only confirmation you will get is with the X-ray diffraction or MNR.

 

[Monique]

Here are 3D modelling tools http://www.bioinformatics.vg/biolinks/bioinformatics/3D%20structure.shtml let us know what works, we'll learn something too

 

[Mike H]

My first rule when dealing with structures: Never trust one until you've examined it for yourself. If it doesn't make sense to you, there's nothing saying you have to accept it.

If you have a structure in .pdb format, a very quick and easy analysis can be done with Swiss PDB Viewer (which should still be free for download, no?). You can do a Ramachandran plot (and some other torsion angles), hydrogen bond lengths, bond lengths, check the surface, and the like. If you're actually working with an experimentally determined structure, there will be other information available (B factors, occupancies, R and R(free), and the like).

If you have unusual features in your predicted or determined structure, one thing that may be informative to do is to do a search of a small molecule database (e.g. Cambridge Structural Database) to see if such features crop up in small molecules. This is also really good for getting a handle on hydrogen bonding, since a number of the structures in the CSD are from neutron diffraction which is generally better in terms of being able to include hydrogens. Although exceptionally high resolution x-ray structures have been able to see hydrogens.

You of course have your own physical/chemical/biological intuition to work with in accessing a structure. How much mileage you get out of this will vary, of course.

You may want to take a look at the CASP competitions (Critical Assessment of techniques for protein Structure Prediction) which have been run for about 10 years now. They're currently on CASP6.

I have a separate but related rant for experimentally determined structures, but since that wasn't asked, I won't go into it.