Support reading and writing intra-residue bonds in PDBx files

[padix-key]

This PR add support for reading/writing intra-residue bonds via the chem_comp_bond category in PDBx files. Files that contain this category were introduced in the PDB NextGen Archive.

[cisert]

I think a pretty common use case might be reading SDF files generates from some cheminformatics software (e.g., RDKit), and then wanting to have the same bond information in cif format.
I've tried reading an RDKit-generated SDF file (attached below, rename to aspirin.sdf), saving it as PDBx (incl. bonds), then reading it again from cif, which didn't seem to keep the bond information:

import biotite.structure.io as strucio
import biotite.structure.io.pdbx as pdbx

arr = strucio.load_structure("aspirin.sdf")
print(arr.bonds.as_array().shape) # (13, 3)

f = pdbx.CIFFile()
pdbx.set_structure(f, arr, include_bonds=True)
f.write("aspirin.cif")

arr2 = strucio.load_structure("aspirin.cif", include_bonds=True)
print(arr2.bonds.as_array().shape) # (0, 3)

Am I missing something?

aspirin.txt

[cisert]

word missing in docstring?

[padix-key]

True, rather an entire sentence. Thanks for spotting this

[padix-key]

I reproduced your problem. There are two issues here:

• The intra-bond parsing code in get_structure() was placed so it only ran for AtomArrayStack. The latest commit fixed this.
• SDF files do neither contain residue nor atom names, so res_name and atom_name are empty in the parsed AtomArray. However chem_comp_bond in PDBx requires these two columns for identifying bonded atoms. Hence, you need to assign them before writing the CIF file:

  import biotite.structure.io as strucio
  import biotite.structure.io.pdbx as pdbx
  
  arr = strucio.load_structure("aspirin.sdf")
  arr.res_name[:] = "ASP"
  # Probably you would want a better atom naming scheme
  arr.atom_name[:] = [f"A{i}" for i in range(arr.array_length())]
  
  f = pdbx.CIFFile()
  pdbx.set_structure(f, arr, include_bonds=True)
  f.write("aspirin_temp.cif")
  
  arr2 = strucio.load_structure("aspirin_temp.cif", include_bonds=True)

[cisert]

Did the code snippet you posted above run for you? Pulling your changes and running locally, I got an error from the final line
TypeError: connect_via_residue_names() got an unexpected keyword argument 'custom_bond_dict'

[padix-key]

Did the code snippet you posted above run for you? Pulling your changes and running locally, I got an error from the final line TypeError: connect_via_residue_names() got an unexpected keyword argument 'custom_bond_dict'

Did you reinstall the package? connect_via_residue_names() is part of a Cython module. Thus, the compiled module will still have the old version, if not explicitly reinstalled.