How In Silico Studies Speed Drug Discovery

In vitro and in vivo studies have well-known weaknesses. In vitro studies tend to over-simplify the dynamic, complicated environments inside living organisms. In vivo studies require significant time, money, and yet still do not correctly simulate the human body.

Enter in silico studies. Computers are able to handle the manipulation of multiple variables simultaneously, and they can do it much more quickly. Drug development has become increasingly reliant on new software to overcome the challenges we’ve traditionally faced.

PROTEIN-LIGAND DOCKING PREDICTIONS

One area in which computer-aided drug design has been particularly helpful is in predicting protein-ligand docking. Traditionally, scientists have had to brute-force solutions by testing thousands of molecules for the desired activity. It’s a slow, resource-consuming process that results in a low success rate of approximately 1%.
 

However, when software is used to predict protein-ligand binding, success rates can reach 50%. In one study, protein-ligand poses were predicted with 90% accuracy1.
 

One challenge that needs to be overcome to further increase accuracy is the simulation of protein flexibility. The 90% accuracy was achieved using one of the only software programs that has partial protein flexibility incorporated into its algorithms.
 

In reality, receptors are flexible, and present programming challenges. Programming each degree of freedom of movement requires significant time, effort, and digital memory. With that being said, however, 90% is much better than 1%.
 

CRITICISMS

Some skeptics aren’t so impressed and are withholding praise until it can be better shown that such models lead to improved drug discovery. Currently, success rate is measured against known protein-ligand interactions. Many feel such retroactive benchmarks of success are biased at best.
 

Still, the area of computer modeling looks promising. There are currently many programs available to aid researchers. Moreover, they are being provided both publicly and free of charge.
 

The lack of shared knowledge among researchers has commonly been cited as a hindrance to rapid scientific development, so the trend towards open-access information is exciting.
 

HUMAN PROTEOME

Another source of excitement is that the human proteome has been thoroughly mapped and over 30,000 proteins have been identified2. Some researchers have used this to their advantage by flipping the protein-ligand search. Instead of comparing many molecules against one receptor, they are comparing one molecule to many receptors. This method has the advantage of eliminating molecules that react with non-target proteins.
 

With all of the developments in protein modeling, it shouldn’t be long before we see more effective drugs entering the market at a higher rate.
 

CLASSES

The Drug Development Process from Concept to Market

ADME, PK/TK & Drug Metabolism in Drug Discovery and Development

Introduction to Molecular Biology Techniques

1. Wang Z, Sun H, Yao X, et al. Comprehensive evaluation of ten docking programs on a diverse set of protein–ligand complexes: the prediction accuracy of sampling power and scoring power. Phys Chem Chem Phys. 2016;18:12964–12975. doi: 10.1039/C6CP01555G

2. http://www.humanproteomemap.org