dominiquesydow commented 3 years ago

Description

At the moment, our ligand data retrieval is a bit too hacky.

We need the (a) complex PDB file that is used to generate the dynophore and (b) the ligand name as used in the PDB.
We extract the ligand atoms using MDAnalysis wit (a/b) to get (c) PDB block and (d) PDB residue IDs.
- Potential problem: We add MDAnalysis as dependency for (c/d) only and we need to know the ligand name used in the PDB to do this.
We need the (e) ligand SMILES in the dynophore JSON file.
We use (c-e) to generate an rdkit molecule by inferring bonds from SMILES.
- Potential problem: This procedure is ambiguous and might not end up to be the same ligand as in the original PBD.

There is a lot of room for crashes :) It would be much better to add

to our dynophore JSON file. That way we can create the ligand directly from the MDL mol block and update the atom IDs with the PDB residue IDs.

Todos

[x] Update dynophores JSON file:
- [x] Add MDL mol block
- [x] Add PDB residue IDs
[x] Update the dynophores.core.Ligand class
[x] Add Ligand class as attribute to Dynophore class
[x] Update backend of view2d module
[x] Update unit tests; rerun notebooks