Deleteing residue from PDB using Biopython library

Question

Using biopython library, I want to remove the residues that are listed in list as follows. This thread (http://pelican.rsvs.ulaval.ca/mediawiki/index.php/Manipulating_PDB_files_using_BioPython) provides an example to remove residue. I have following code to remove residues

 residue_ids_to_remove = [105, 5, 8, 10, 25, 48]
 structure = pdbparser.get_structure("3chy", "./3chy.pdb")
 first_model = structure[0]
 for chain in first_model:
     for residue in chain:
         id = residue.id
         if id[1] in residue_ids_to_remove:
             chain.detach_child(id[1])
 modified_first_model = first_model 

But this code did not work and raised the error

def detach_child(self, id):
    "Remove a child."
    child=self.child_dict[id]
    KeyError: '105'

What's wrong with this code?

Alternatively, I can do using accept_residue() and write it in PDB. I don't want to follow like this becauseI want to do it while in memory for further processing.

Answer 1

Biopython cannot find the key in the inner dictionary to the chain, because you're supplying a random key. The dict looks like this:

child_dict = {(' ', 5, ' '): <Residue HOH het=W resseq=5 icode= >,
              (' ', 6, ' '): <Residue HOH het=W resseq=6 icode= >,
              (' ', 7, ' '): <Residue HOH het=W resseq=7 icode= >}

That is: using tuples as dict keys. You can see the dict doing print chain.child_dict.

Once you know this, the error/solution is clear. Pass a valid key to detach_child, namely remove the [1]:

   if id[1] in residue_ids_to_remove:
       chain.detach_child(id)

The right way

Detach the children from the chain level, and do not cycle the residues directly:

for chain in first model:
    for id in residue_ids_to_remove:
        chain.detach_child((' ', id, ' '))

Or with list comprehension:

for chain in first_model:
    [chain.detach_child((' ', id, ' ')) for id in residue_ids_to_remove]