David Koes
David Koes
It treats them as part of the receptor. By default, the AutoDock Vina scoring function is used for search and refinement and the CNN scoring function for ranking and scoring.
By default, Vina is used every except for the final ranking, but this can be changed with the `cnn_scoring` argument.
Both the grid and the "equational" scoring function are the same function - the Autodock Vina function which you can get from the Vina paper (https://doi.org/10.1002/jcc.21334). The grid approach is...
Empirical scoring doesn’t use GPUs.
It isn't using the GPU for calculations even if it may have initialized some data structures on it. To avoid the GPU altogether set CUDA_VISIBLE_DEVICES=
CNN_VS is CNNaffinity*CNN_score CNNaffinity is a binding affinity prediction in pK units (e.g. 9 is 1 nM) CNNscore is a predicted probability that the pose is good (
No, but is easy enough to add them back with your favorite cheminformatics toolkit.
The point of docking is to generate a new geometry. If you have examples that you think are problematic please share them.
I find hydrogen atom geometry uninteresting.