Limitations on the max size of a molecule?

[iGulitch]

At the web page of the web-based Ligpargen service, it is written "Maximum ligand size allowed is 200 atoms". This is understandable. However, I thought naively that the stand alone version of Ligpargen did not have this limitation. And this is not the case unfortunately. Namely, too big molecule submitted to Ligpargen leads to an error :

Traceback (most recent call last):
  File "/opt/conda/envs/ligpargen/bin/ligpargen", line 33, in <module>
    sys.exit(load_entry_point('LigPargen', 'console_scripts', 'ligpargen')())
  File "/ligpargen/ligpargen/ligpargen.py", line 504, in main
    molname=args.molname, workdir= args.path, debug= args.debug)
  File "<string>", line 17, in __init__
  File "/ligpargen/ligpargen/ligpargen.py", line 192, in __post_init__
    moleculeA = Molecule.fromBOSS(zmatName, outFile, pdbFile, moleculeA.shiftX, moleculeA.shiftY, moleculeA.shiftZ)
  File "/ligpargen/ligpargen/topology/Molecule.py", line 159, in fromBOSS
    atoms, numberOfStructuralDummyAtoms = cls._getAtoms(cls, zmatData, pdbfile)
  File "/ligpargen/ligpargen/topology/Molecule.py", line 496, in _getAtoms
    atomsLines = re.search(r'BOSS(.*?)Geometry', zmatData, re.DOTALL).group().splitlines()[1:-1]
AttributeError: 'NoneType' object has no attribute 'group'

I had to dig into the Ligpargen code to figure out that the failure happens due at that stage, where BOSS is involved. Having read BOSS manual, I figured out that there were some limitations on the molecule size, but I didn't get them clearly.

May I ask to clarify the internal limits of BOSS and of Ligpargen [ if there are any ] on the maximum number of atoms in a molecule?

[atomrq]

I got the same problem. The Ligpargen offlien version can not process large molecule with atoms more than 200. I also wonder is is a rigrous limitation in the code?

[iGulitch]

@atomrq , I couldn't find any limitations embedded in the Ligpargen code itself. What I figured is the following :

1. Ligpargen hinges on RDKit and OpenBabel when it comes to molecule processing, e.g. reconstruction of a molecule based on the SMILES sequence. For instance, check utilities.py. RDKit and OpenBabel themselves might struggle with a large molecules processing.
2. Ligpargen calls BOSS at some point, which has its own internal limitations [ read the BOSS manual ] .

I'm looking forwards to seeing the reply from the developer [-s] :)

[Isra3l]

Hey everyone,

BOSS has this limitation, and as the source code is not open, there is no way to use molecules bigger than 200 atoms.

[iGulitch]

Dear @Isra3l ,
thanks for the reply!

In fact, 200 atoms isn't the precise number. I managed to parametrise a polymer molecule with 247 or 257 atoms, but failed to do so with the one that consisted of ~270 atoms. On the other hand, BOSS manual itself, that isn't entirely clear unfortunately, mentions some limit that is about 450 objects.