The main focus of this project is to use machine learning to improve the ability to find correct poses for protein-ligand binding, which will aid in future drug design. Given 795 protein-ligand crystal structures, we will generate 100 “decoy” ligand poses per protein. These simulations are created using Schrodinger’s Glide docking package. After generating the decoys, a scoring function will be applied to determine the correct ligand pose. We will then employ random forest machine learning to optimize the ability to predict poses that are close to the native crystal structure. This plays a major role in drug design because the user can take the initial structure and find a molecule that will fit into the pocket, for example, a molecule that inhibits schizophrenia by docking to serotonin receptors. In the future, we hope that this algorithm can optimize predictions for other data sets.