Ever heard of the term bioinformatics? It’s an interdisciplinary field of science that combines different topics such as biology, computer science, mathematics and statistics. In a world where data is being generated at a faster rate than we can process, bioinformatics is used to analyze massive amounts of data and make sense of it all. We take a look at how bioinformatics is making a huge impact in drug discovery and design, by focusing on the field of molecular docking in virtual screening of compounds.
In Vivo, In Vitro, In Silico
You may have already heard of in vitro and in vivo; techniques that take place outside a living organism, and those performed within a living organism. As the lesser-known third category of study, in silico techniques come under the umbrella of bioinformatics research. The expression in silico was first used in 1989, meaning ‘performed on a computer or via computer simulation1.
When it comes to drug design, it is intuitive to see why in silico methods are such a good idea: The stages that precede the design of a new drug are both costly and time-consuming. The whole process can take from 12 to 15 years and cost a few millions of dollars, but in silico studies have been seen to both speed up the discovery rate and reduce (although not eliminate!) the need for expensive lab work.
In silico research is performed during stage 1 – the drug discovery and design process – comprising many important methods such as homology/comparative modeling, molecular docking, virtual high-throughput screening, quantitative structure-activity relationship methods (QSAR), conformational analysis, the list goes on!
In this article we will focus on molecular docking since the author has personally worked for some time with the technique – modeling antibodies is tougher than it might seem!
First off, let’s try to understand the concept of docking and its implications. Molecular docking a type of bioinformatic modeling, an essential tool in structural molecular biology and in drug design. The purpose of using this technique is to predict the most likely ‘binding scenarios’ between a protein and a ligand of known three-dimensional structure4,5.
The diagram below shows a simplified depiction of how the docking procedure can influence and empower drug design. However, while this technique might seem to be able to reveal potential drugs rather easily, in silico methods and simulations are definitely not a substitute for good ol’ laboratory assays!
The entire process is centered around using software to generate heaps of ligand-protein conformations and selecting the ones thought to bind most strongly. The binding affinity of the ligand for the receptor is predicted by simulation software (there is a huge variety for both academic and commercial purposes), using mathematical equations known as scoring functions.
‘Scoring functions’ are able to predict to a certain extent the ‘binding free energy’ between the ligand and the target, with the aid of ‘sampling algorithms’ that attempt to reproduce the actual binding event as close as possible5,6.
The importance of the sampling algorithms used in docking procedures is evident when the sheer number of possible conformations between the ligand and protein is taken into consideration; factors that influence the binding energy include the huge number of translation, rotational and conformational degrees of freedom. It turns out that this is actually way too much computation… even for a computer (the irony!) In order to refine the search – saving time and costs – these algorithms are put in place in order to remove improbable events.
Now that we’ve removed nonsense conformations that are unlikely to exist, we can sift through the manageable (but still enormous) pool of binding scenarios using the aforementioned scoring functions. Rather than using precise calculations, these functions estimate the binding energies of many different conformations in a reasonable amount of time, albeit by sacrificing some degree of accuracy. The different scoring functions used are highlighted below5,6:
Classical force-field functions
They assess the binding energy by calculating the sum of non-bonded interactions such as electrostatic interactions and Van der Waals forces. Some extensions of these calculations include other relevant aspects such as hydrogen bonds or entropy.
Empirical scoring functions use known binding affinities of protein-ligand complexes to perform multiple linear regression analyses. The values generated by this statistical model are then used as coefficients to adjust the equation in general. These ‘coefficients’ rely on the chosen dataset as well as the type of software used, making it common for differing results to be generated.
These functions use statistical analysis of ligand-protein crystal structures to obtain the measurement of the distance between them; it makes the assumption that an interaction that looks favorable will lead to activity. Its advantage lies in its computational simplicity, with a drawback being some interactions might not be represented in the available database of crystal structures. The operator also has to choose a protein set to ‘train’ the program with, which also means differing results.
This introduction provides a little bit of background behind the whole process of molecular docking; showing how software can calculate how well your drug can fit with the target protein.
Of course, there are other factors to take into consideration, such as how to treat individual receptors and ligands – rigid components that fit a certain position (an over-simplified version of the reality and because of that, not very popular nowadays) or as flexible structures (sounds more logical right?)
I could keep going on and on with the whole process behind molecular docking, which is a big part of computing sciences, biology and physics, but I feel like I can stop here. Essentially, molecular docking is a highly versatile tool that provides a nice push to an overly-slow industry, making the development of new drugs a bit less tedious and definitely more interesting in terms of understanding how proteins bind!
- Mpkb.org. (2018). Differences between in vitro, in vivo, and in silico studies (MPKB). [online] Available at: https://mpkb.org/home/patients/assessing_literature/in_vitro_st/span>
- Yourgenome.org. (2018). How are drugs designed and developed?. [online] Available at: https://www.yourgenome.org/facts/how-are-drugs-designed-and-developed
- de Ruyck, J., Brysbaert, G., Blossey, R. and Lensink, M. (2016). Molecular docking as a popular tool in drug design, an in silico travel. Advances and Applications in Bioinformatics and Chemistry, Volume 9, pp.1-11.
- Gao, Q., Yang, L. and Zhu, Y. (2010). Pharmacophore Based Drug Design Approach as a Practical Process in Drug Discovery. Current Computer Aided-Drug Design, 6(1), pp.37-49.
- Hughes, J., Rees, S., Kalindjian, S. and Philpott, K. (2011). Principles of early drug discovery. British Journal of Pharmacology, 162(6), pp.1239-1249.
- Meng, X., Zhang, H., Mezei, M. and Cui, M. (2011). Molecular Docking: A Powerful Approach for Structure-Based Drug Discovery. Current Computer Aided-Drug Design, 7(2), pp.146-157.