Principle of MolEV
In the generation of the molecule, we divided it mainly into two parts called molgraph and prograph, respectively, in Molgraph, we extracted the dihedral angle of the molecule With two geometric angles α, β, we set four atoms of the molecule into one set (α, β, γ) arrays and mapped to RGB triple primaries
Model Details
Framework about predicting the protein pocket
We built the neural network prestige with the CNN as the main body of predicted protein binding pockets.A four-channel grid protein descriptor is constructed based on ligand site (ligsite), L-J potential, and Coulomb force, which can be used for protein binding site predictio
Research Details
