Full Structure Of Protein Sequence

Question

[Sheila]

Hello, I have some protein sequences and looking for structures. Doing a blast search in PDB gives many hits but they do not cover full sequence. For an example doing blast search in PDB for p53 gives many hits but none of them covers full sequence. The human p53 sequence have length 393 amino acids (in uniprot) while in PDB maximum alignment Length is 219 (only 55% of original sequence). So what about the other 45% of sequence? I understand that PDB and uniprot have different approach for protein information. But I am specifically looking for the full length of the sequence.

Is there any database (based on homology modeling or other methods) which have full structure of protein sequences (It should be downloadable/accessible form script) ???

[Sheila]

Is there any Database which can be download and have full protein structure ??

Answer 1

[NPalopoli]

You should notice that it is very common to find PDB files that do not span the whole length of the actual protein. This may be because the experimentalists have focused on just one domain of a multidomain protein, or because they have not been able to solve the positions of some of its atoms (perhaps, as João Rodrigues' suggested before me, due to the disorder properties of the sequence).

If you need to have a structural model of your entire protein, you could try a whole fold and function assignment using the FFAS server against the PDB database. It performs HMM-HMM searches which should be more sensitive than most other alternatives. FFAS is accessible at http://ffas.ljcrf.edu/ffas-cgi/cgi/ffas.pl

Once you recover PDB IDs that match to the different parts of your protein, you may try to model them with MODELLER.

You shouldn'd have much trouble trying this but please don't hesitate to look for an answer here or let me know in private if you need my help.

[Sheila]

Is there any pre-calculated Database which can be download??

[NPalopoli]

I am not entirely sure what you are asking for. If you just need a database of full proteins then the ProteinDataBank is the place to go. If your protein is not deposited there, or just a part of it is, then you will need to find different template PDBs that cover the total length of your protein and perform a multi-template homology modelling procedure with MODELLER as described in http://salilab.org/modeller/tutorial/advanced.html

Answer 2

[João Rodrigues]

If you go to Uniprot, near the bottom you have the structural information pertaining that particular entry. You can get several "pieces" of the protein and then superimpose them (in Pymol f example), giving you a complete model. If there are still parts missing, homology modelling (MODELLER for scripting is good, MODBASE for precomputed non-optimal models).could be used to fill in these blanks.

You should also look at the disorder properties of the sequence that is missing, maybe that is why you can't find a structure and building an homology model will likely impose secondary structure features that are unnatural.