This PR is a draft to self-consistently parameterize protein-ligand systems with espaloma-0.3.0 which was built on top of perses-0.10.1. Note that this PR is not ready to be merged to the latest release but intended to keep track of how one could self-consistently parameterize a protein-ligand systems with espaloma-0.3.0.
Motivation and context
Current implementation of perses cannot parameterize the protein-ligand system self-consistently with espaloma.
One work around is to parameterize the system with the current implementation (protein: Amber, ligand: espaloma), and use the solvated system to regenerate the protein with EspalomaTemplateGenerator implemented in openmmforcefields.generators.
A more simple way is to handle the protein as a single residue openff.topology.Molecule object and append it with the other small molecules, Then load the list of Molecules into openmmforcefields.generators.SystemGenerator. However, this raised some issues described here.
Thanks for this @kntkb ! This will help ups come up with the changes we need to be doing in our API to be able to parametrize receptors/proteins as well.
Description
This PR is a draft to self-consistently parameterize protein-ligand systems with
espaloma-0.3.0
which was built on top ofperses-0.10.1
. Note that this PR is not ready to be merged to the latest release but intended to keep track of how one could self-consistently parameterize a protein-ligand systems withespaloma-0.3.0
.Motivation and context
Current implementation of perses cannot parameterize the protein-ligand system self-consistently with
espaloma
. One work around is to parameterize the system with the current implementation (protein: Amber, ligand: espaloma), and use the solvated system to regenerate the protein withEspalomaTemplateGenerator
implemented inopenmmforcefields.generators
.A more simple way is to handle the protein as a single residue
openff.topology.Molecule
object and append it with the other small molecules, Then load the list ofMolecules
intoopenmmforcefields.generators.SystemGenerator
. However, this raised some issues described here.How has this been tested?
This version was used to run the custom PL-benchmark described here.