I am going to describle the problem completely. First I should say that there are two kinds of interactions: transient and stable interactions. We are supposed to work on stable interaction like interactions between two monomers in a heterodimer. In a heterodimer there are two chains in which there are some residues between monomers which take part in the interaction of monomers and building the structure of heterodimer and also there are other residues interacting inside each monomer for building the structure of that monomer. There are some PDB IDs related to heterodimers in PDB. Now imagine that we are trying to show which residues are physically interacting between two monomers in a heterodimer and then we are trying two show the site of these physically interacting residues in the sequence related to that special structure. For that, we need to find the sequence related to that structure and we can download it from PDB. But if we want to make a multiple sequence alignment of that sequence, first we need to find the refseq sequence of that and then we must blast the sequence and find homologues of that sequence. Here there are some challenges: first, we must find residues interacting in the structure of heterodimer between two monomers. Second, we must find the sequences of monomer chains of the heterodimer and map the interacting residues in the structure to the sequence. Third, we know that when a structure is solved and its monomers are sequenced, may be they are not able to completely sequence the heterodimer and residue numbers are maybe different from the related refseq sequence. I would like to know how can we find the related refseq sequence (or sequences, in the case of heterodimer) and then how can we map the physically interacting residues in the structure of PDB ID to refseq sequence (finding the sites of these interacting residues in that refseq sequence) regarding three challenges mentioned above. Imagine that I have 100 PDB IDs of 100 nonredundant heterodimers and I would like to find and download the structures, sequence and finally refseq sequence. What should I do?
My reputation isn't high enough to make a comment, but have you considered a modelling software such as PyMol or Chimera? You can execute a command line script with softwares such as these and these have the functionality to predict H-bonds.
In the PDB files there is information regarding the GenBank and UNIPROT info of the protein, and you can find more information here:
If you want RefSeq ID, you can use GenBank info you extract and convert by piping into a biomaRt (Bioconducor) script, although maybe there is a more direct way.
As for finding which residues in the sequence are interacting, I'm fairly certain that the software packages like PyMol or Chimera will give you information about the interacting residues.
From what I understand, you will want to filter out only the residues that interact and are on different chains (e.g, Chain A vs. Chain B). This process will require you to work with the syntax of the different software packages when you are scripting. For example, Chimera has a basic primer here:
Note that I only gave a reference to Chimera because I am more comfortable with that then other packages and there is likely a more efficient way to do this. With Chimera, you can find the position within the chain of residues participating in the interaction and then you'd store these in a file to do whatever you want later.
I hope this helps.