Computational Drug Discovery: Technologies and Methods

Previously, we discussed in part 1 of this series the growing significance of computational methods in drug discovery. Traditional drug discovery typically uses tools such as high-throughput screening in order to find novel therapeutics: through screening a large number of molecules in order to find ones that have the right biological response. The assays that are used to screen these molecules can be time consuming, and often times there are only a few hits that develop into usable drug leads. Computational drug discovery not only allows for shorter screening times but it can decrease the number of compounds to test by removing those that are predicted to be inactive and prioritize those that are predicted to be active.

One aspect of the drug development pipeline that is aided by computational methods is the target identification. Scientists can search widely available databases, such as the Database of Essential Genes and the National Center for Biotechnology Information Website, that contain information on various proteins that are vital in cell function that can then be targeted towards drug development.² In fact, Scientist has 78 vendors that offer bioinformatics services, allowing for specific requests for a particular bioinformatic analysis.

After deciding on a particular protein as a drug target, computational drug discovery can aid in discovery of a lead and in lead optimization. These leads can potentially become further developed into usable drugs. There are two typical ways to screen for a potential lead, through structure-based methods or ligand-based methods. Structure-based computational drug discovery requires knowledge of the protein structure that the drug is potentially targeting, while the ligand-based method is generally used when little or no structural information is available. Often times, both methods are used to complement each other. Scientist also offers 29 vendors that provide these computational protein chemistry services, which can aid in the theoretical testing of potential leads and targets in the drug discovery pipeline.

There are several molecular modeling simulations software programs available, such as AutoDock, Zodiac, PyMOL, Discovery Studio, and Ascalaph. These programs allow for the 3D visualization of the interaction between the protein target and the potential compound that could be a drug lead. There are also various programs that allow for the lead optimization through the changing of various functional groups of a compound.³ These computational methods for lead identification and optimization allow for extremely quick high throughput screening at very low cost. Additionally, researchers using these computational methods will either use a pre-set scoring function or create a scoring function to determine which of the hits are usable as viable drug leads and which can be thrown out. Once these viable drug leads are determined, researchers can then physically assay them and go through the subsequent steps in the drug discovery pipeline.

Now that we have some examples of the technologies and methods used for computational drug discovery, tune in for next week’s part 3 in this blog series where we will discuss the increased applications of computational drug discovery!