2025-04-09 Meeting notes

Use RDKit for “internal coordinates” ?

MT

• https://www.rdkit.org/docs/source/rdkit.Chem.rdmolops.html#rdkit.Chem.rdmolops.Get3DDistanceMatrix

• Get 3D distance matrix with rdkit #RDKit

• LW: other RDKit functions:

  • https://www.rdkit.org/docs/source/rdkit.Chem.rdMolTransforms.html#rdkit.Chem.rdMolTransforms.GetAngleDeg

  • LW: I’m not sure if you can calculate impropers easily. I use MDAnalysis for that. FYI, MDA is now LGPL and can be imported and used non-infectiously.

Danny Cole torsion constraint/restraint opinions

MT

• Resolved or not? Couldn’t find ZenHub ticket

  • LW: Finlay did a look at RMSD comparisons in restraining non-dihedral atoms. We didn’t see much difference in bulk comparisons over the dataset, on the order of <0.01 A going from restrained to unrestrained. For comparison we use 0.4A as a threshold for ddEs. There were some outliers he was going to look at.

Software-dependent minimized RMSDs?

• Finlay did a deep dive and found that OpenMM & geomeTRIC optimized to different RMSDs

  • about 3/4 through recording linked in 2025-04-09 Cole Lab Check-In Meeting Notes

  • Slide 17: https://newcastle-my.sharepoint.com/:p:/g/personal/nfc78_newcastle_ac_uk/EYhL9U9wU0RBv1O-gQZQ5hsBqJl4z5b9hUTLXgSYD-wBNQ?rtime=pwM1uqR33Ug

• These differences separate from whether or not heavy atoms are included

  • IIRC Lexie looked at something similar and it has to do with both the minimizer used and the coordinate system used