Using preoptimized ligand geometries instead of smiles strings as input

[nhadler]

Hi! It wasn't clear from the documentation in the README, but can I provide ligands as any other form than smiles strings? Possibly as an optimized mol or xyz file?

Using smiles strings as the sole possible input is problematic if you need to encode specific (axial for example) chirality that cannot be encoded with smiles strings or ligands with dative bonds like ferrocenyl phosphine ligands.

[mgt16-LANL]

Hi @nhadler - this is an interesting proposition/idea. I had started to implement this, but didn't find many use cases. Happy to add when I have a chance to implement!

In the meantime, I am wondering about your two examples, as both should be supported already:

Manual Pinning of Symmetries:

from architector import (build_complex,
                        view_structures) 
# The major routines used in architector
from architector.geometries import octahedral
# Explicitly defining octahedral environment
print([(i,x) for i,x in enumerate(octahedral)])
inputDict = {'core':{'metal':'Fe','coordList':octahedral},
           'ligands':[{'smiles':'p1ccccc1-c2ccccn2', # P and O should be trans
                       'coordList': [[0,2],[11,4]]}, 
                       {'smiles':'CC(=CC(=O)C)[S-]', # N and S should be trans
                        'coordList':[[4,3],[6,5]]},
                       {'smiles':'O', 'coordList':[[0,1]]}, # And water at the -z site
                       {'smiles':'[Cl-]','coordList':[[0,0]]}], # And chloride at the +z site
           'parameters':{'metal_ox':3}} # Add oxidation state here.
fe_octahedral_out = build_complex(inputDict) # Now just build using the dictionary!

What I think is a ferrocenyl phosphine ligand:

pcp = {'smiles':'[CH]1[CH][CH][C-]([CH]1)Pc2ccccc2', 
      'coordList':[0,1,2,3,4],
      'ligType':'sandwich'}
out = build_complex({
   'core':{'metal':'Fe','coreType':'octahedral'},
   'ligands':['cp',pcp],
   'parameters':{}
})
view_structures(out)

[nhadler]

Thanks! I wasn't sure how to generate the ferrocene ligands, that's a big help!

Though being able to provide ligands as mol files would be a massive help for ligands like binap/segphos/etc where ensuring the ligand's absolute configuration is critical; especially for applications in predicting things like enantioselectivity. I would be really appreciative if that could be implemented!

[mgt16-LANL]

Awesome. I hope to have this functionality built out on the dev branch (Secondary_Solvation_Shell) soon (hopefully end of august).