i finds thousands of files in protein data bank,in which many chains have missing residues it creates problem while calculating phi & psi
also another problems:- there are some residues in chain ,where one of atom type e.g.N, CA, C, O (which are essential for phi/psi calculations) are not given/absent
what should i do? any logic/program/server......
The way I have dealt with this problem is to use Modeller (http://salilab.org/modeller/) to "impute" missing residues and atoms. It is not a trivial task (you have to write code), but this should get you started:
http://salilab.org/archives/modeller_usage/2002/msg00188.html
If PDB structure has a missing residue you just don't calculate phi/psi angles for neighboring residues...
In a case you want to derive some statistics about phi & psi angles in protein structures, one direct solution to your problem is to use high resolution structures only. Such structures should have much less missing atoms. For that kind of analysis you should use high resolution structures only anyway as low resolution ones will have more uncertainties/errors.
For cases where a residue is not missing but some of its atoms are missing, perhaps you can add missing atoms using Modeller's complete_pdb() (http://salilab.org/modeller/manual/node432.html). This will not work however for adding whole missing residues.
BTW. the missing residue information is included in pdbx_poly_seq_scheme field in mmCIF as questions marks . Also, this information is directly accessible through SEQATOMS database
Anyone who is having the same problem, Please use http://lorentz.dynstr.pasteur.fr
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
The post from that modeller_usage mail list uses outdated modeller interface. Besides adding missing residues to PDB structures by modeling (de novo) make sense only in some applications. If 'true' phi/psi angles from PDB are needed, for example for statistical analysis, modeling by Modeller does not make sense.