Designing drugs on the London Underground

When you think of drugs and the London Underground, you probably don’t get a completely wholesome, let alone scientific image. I am going to tell you how we can use systems such as the London Underground to improve the design of new drugs.

The London Underground map is an example of a mathematical network, or graph. This stems from the growing field of ‘graph theory’. It is made up of individual points, the stations, joined together by lines, the train tracks. But, we can copy this idea in biology. We can make a network like this, where the stations are individual proteins, and the train tracks between them are interactions between these proteins. This is known as systems biology.

But why would we want to do this? Traditionally, biology has looked at very specific things, like individual genes or proteins in great detail. But imagine you were to study a car, you wouldn’t just look at the engine or brakes in detail, you would look at how all of these components come together to make the whole vehicle. This is the approach we now need to take with biology, we need to start looking at the bigger picture. Building a biological network, like the London Underground, allows us to see this bigger picture.

Quite amazingly, when we do this, we see patterns arising. These patterns are present not only in molecular networks of proteins but also in networks of biological food webs or even social networks. Viewing these networks as a whole quickly allows us to see the key components (as these will have the most connections with other components), and which components are closely related, as they will interact with similar things, affecting similar cellular processes. This may not necessarily be obvious when only viewing the individual parts.

Credit: Eichelbaum et al., 2012, Nature Biotechnology

This picture shows an example of a biological network. Each circle represents a protein and the lines between them represent their interactions. It is clear to see that closely related proteins have more interactions between them and are grouped together. This allows biologists to see patterns and predict the functions of newly discovered proteins more easily, simply by looking at the other proteins it interacts with.

Now, and this is the exciting part, we can use these networks for the design of new drugs. Currently only 1 in 5000 drugs that enters pre-clinical testing actually makes it to market. Any drugs that fail cost the industry millions of pounds, usually failing because of unexpected side effects. Using systems biology, we can see from our network which components would make suitable drug targets and which would make poor targets, and, by looking at the interactions that exist, we can use it to predict any whole-system side effects of these drugs. This would not be possible by looking at the individual parts, and only becomes apparent when we view the effects on the system as a whole.

What’s more, using mathematical modelling, we can add experimental data to our network and turn it into a sort of ‘virtual cell’ on a computer. We can then use this virtual cell to make predictions about what would happen if we altered the network with various drugs, giving much more accurate predictions of which new potential drugs should be developed.

So by using systems biology, we can more accurately predict the positive and negative effects of a drug, before it has even been designed, before we have even been in the lab! This will lead to cheaper and more efficient drug design. But, this is only becomes possible when we can take a step back and view the bigger picture, something which we are finally starting to do.

Tags: , , , ,

2 Responses to “Designing drugs on the London Underground”

  1. Ram Samudrala May 31, 2015 at 12:31 pm #

    We have been doing this for more than a decade now and have developed a computational drug discovery platform that looks at the interactions between 3733 human use drugs and 48,278 protein structures (in v1, currently we’ve over 60K proteins) to determine drug behaviour at the proteomic level. We’ve successfully used it in 12 studies covering 9 diseases in vitro, in vivo (mouse), and we’re trying to do some clinical studies now. We’ve published dozens of papers related to this ( and more generally for the basic science behind this, – PDFs of the papers referred to below are available there as well as all the others).

    Our two recent papers were published in Drug Discovery Today and Mini Reviews in Medicinal Chemistry:

    and some of our first work on this was published in 2004 in the Journal of the American Medical Association and in 2008 in Trends in Pharmacological Sciences:

    This project was funded in part by a 2010 NIH Director’s Pioneer Award, which is one of the most prestigious grants that NIH gives out.


  2. Elizabeth June 9, 2015 at 7:52 pm #

    This is a really clear description Ben, I enjoyed reading it.

Leave a Reply