search

openforcefield / openff-benchmark

Comparison benchmarks between public force fields and Open Force Field Initiative force fields
MIT License
10 stars 2 forks source link

Post season 1 refactor: one vision #88

Open dotsdl opened 3 years ago

dotsdl commented 3 years ago

A vision for openff-benchmark is that it offers the following:

  1. An execution pathway for applying QM, MM, and ML potentials to sets of molecules for the purposes of comparison via a variety of metrics. Geometry optimizations and torsiondrives are two such execution pathways.
  2. A library of metrics relevant for benchmarking, including e.g. per-moiety RMSDs; RMSD- and energy-based comparison of conformers from different MM forcefields, QM calculations, ML potentials; FF coverage.
  3. Data structures for handling execution in (1) and utilizing the metrics in (2) in an understandable and reliable way.
  4. A command-line interface for applying all of the above on serialized collections of molecules, such as a collection of SDFs.

This vision is one option for the upcoming refactor, and is articulated here from recent group discussions.