Computer-aided drug design is an indispensable process in modern drug discovery. The CADD methods can be roughly classified into two categories based on whether biologic targets are used in the methods or not. In most cases, a drug is an organic small molecule that activates or inhibits the function of a bio-molecule (such as a protein), which in turn results in a therapeutic benefit to the patients. In the most basic sense, drug design involves the design of small molecules that are complementary in shape and charge to the bio-molecular target with which they interact and therefore will bind to it. 

A drug design method that relies on the knowledge of the three-dimensional structure of the bio-molecular target is known as structure-based drug design, such as molecular docking. While a method need no information on the bio-molecular target at all belongs to the ligand-based drug design, such as pharmacophore modeling. In ligand-based drug design, the ligands are assumed to bind to the bio-molecular target with the same or similar mechanism. 

The basic rational drug design procedure is summarized in the following figure: 

In molecular docking, scoring functions are fast approximate mathematical methods used to predict the strength of the non-covalent interaction (also referred to as binding affinity) between the two molecules after the ligand (usually a small organic molecule) have been docked to the receptor (such as protein and DNA). 

A scoring function in molecular docking falls into one of the three categories: molecular mechanical force field (MMFF)-based, empirical-based or knowledge-based. We are interesting to develop high quality MMFF-based scoring functions to accurately predict the binding affinity, not only considering the van der Waals and electrostatic interactions, but also the solvation contribution. The desolvation energies of the ligand and of the protein are taken into account by using implicit solvation methods of GBSA and PBSA. 

Though high-throughput screening (HTS) is routinely performed in pharmaceutical companies to identify drug leads, virtual screening (VS), a complementary approach to HTS is indispensable in modern drug discovery owing to its unique advantages over HTS, such as its negligible operating cost compared to a HTS experiment. The low operating cost of VS enables us to explore a much larger chemical space of pharmaceutical molecules for a target. Moreover, VS in most cases produces target-relevant hits. In contrast, HTS may produce many hits due to nonspecific bindings. The mission of this core facility is to provide virtual screening service to the researchers who want to develop small molecules to inhibit their protein or nucleic acid targets. Virtual screening will be performed using a hierarchical virtual screening strategy for the ZINC database, a famous database of commercially-available compounds for virtual screening. The possible VS filters include molecular docking, binding free energy calculated with MM/GBSA and MM/PBSA, pharmacophore models, 2D/3D QSAR models, substructure or 2D/3D fingerprints, etc. A good balance between accuracy and efficiency can be achieved by this hierarchical strategy since the cheap filters are applied at the earlier stages, and more advanced filters are applied only after the tree is pruned (See Figure 2). The exact filters for a specific VS study depend on the already known information on the target and inhibitors. As ADME-Tox (Absorption, Distribution, Metabolism, Excretion, and Toxicity) is a major reason that leads to attrition of drug candidates, the ADME-Tox and drug-likeness profiles of VS hits will be predicted. The promising hits with good ADME-Tox and drug-likeness profiles will be reported to the users for further testing. 

ADME-Tox (Absorption, Distribution, Metabolism, Excretion, and Toxicity) is a major reason that leads to attrition of drug candidates. 

We have developed a set of in silico models for modeling many ADME-Tox properties including human intestine absorption (HIA), human oral bio-availability, aqueous solubility, protein binding, urinary excretion, blood-brain partitioning, AUC etc. 

We have also developed a novel algorithm to analyze drugs' building blocks. A web toolkit is being developed to predict the ADME-Tox properties with those models for an arbitrary molecule.