Catch divide-by-zero error in energy_ratio.py

[amorehead]

• Catches a rare "divide-by-zero" error in energy_ratio.py.

[maabuu]

Thank you for the PR. Could you add a unit test for when this used to fail? Alternatively if you share examples for molecules where this was a problem, I'll add it. I am curious if there are examples other than single atom molecules.

[amorehead]

Hi, @maabuu. Attached is the batch of input complexes I tried providing to PoseBusters using the redock config. I have been trying to evaluate all 106 of these inputs for around 24 hours, and the bust_table pipeline still seems to be running.

Code for running PoseBusters on these inputs:

buster = PoseBusters(
    config="redock", top_n=None
)
bust_results = buster.bust_table(mol_table, full_report=True)

PoseBusters_Frozen_Bug_Inputs.zip

[maabuu]

Release v0.2.11 now includes this fix. This handles molecules where the UFF returns 0 for the energy. This was a problem for molecules that only contains atoms and no bonds.

[amorehead]

Update: the ZIP archive of input examples I previously uploaded was erroneously not segmented into individual molecular fragments. I have attached below the corrected version of these inputs as a new ZIP archive.
PoseBusters_Large_Examples.zip