|
# LigBoundConf 2.0 |
|
|
|
LigBoundConf is a set of high-quality drug-like bound ligand structures in the Protein Data Bank (PDB). The original set contains three sets |
|
of structures: "raw” structures taken directly from the PDB, "bound” structures, obtained by performing a local minimization of the ligand |
|
in the presence of the protein pocket with an empirical force field, and “global minimum” structures, obtained by finding the lowest energy |
|
conformation of the ligand with an empirical force field and implicit water solvation model. |
|
|
|
LigBoundConf 2.0 is a neutral subset of LigBoundConf in which we have recalculated the "bound" and "global minimum" structures in vaccuo |
|
without an implicit solvent. To obtain the "bound” structures we sought to relieve any artificial high-energy artifacts from the deposited PDB |
|
structures. We did this by running a local minimization with the semi-empirical quantum chemistry method GFN2-xTB using the “crude” |
|
convergence criteria. This ensured that artifacts were removed without significantly altering the bound pose of the ligand. We then performed |
|
single point energy (SPE) density fucntional theory (DFT) calculations at the ωB97M-D3(BJ)/def2-TZVPPD level of theory. |
|
|
|
To obtain the "global minimum” structures, we used the CREST enumeration method with 50 conformers, which uses the same GFN2-xTB energy |
|
function to search for low energy gas phase structures. We took the lowest energy structure as the ``global minimum” conformation and |
|
performed a SPE DFT calculation at the same ωB97M-D3(BJ)/def2-TZVPPD level of theory. |
|
|
|
The LigBoundConf 2.0 neutral dataset ended up with 3132 ligands. |
|
|
|
This dataset is used to validate the StrainRelief method presented in here ([code](https://github.com/prescient-design/StrainRelief)). |
|
|
|
--- |
|
license: apache-2.0 |
|
--- |
|
|