UK +44 (0)1865 849841
Malaysia +603 2731 9335

The benefits of network-driven drug discovery

e-Therapeutics, an Oxford-based company, is using its network-driven drug discovery (NDD) technology to embrace the inherent complexity of biology, providing a novel and productive method for the discovery of new medicines.

Jonny Wray is the head of Discovery Informatics at e-Therapeutics. Trained as a computational neuroscientist, he is responsible for building the NDD technology. We spoke to him ahead of his presentation at the Global Pharma R&D Informatics & AI Congress.

Tell me a bit about network-driven drug discovery?

Network-driven drug discovery is an application of network biology concepts to the drug discovery problem.

Cellular phenotype, both normal and pathological, doesn’t arise from the behaviour of individual molecules. Rather, it arises from the co-ordinated interaction of multiple molecules within the cell – in other words phenotype emerges from the synergy of networks of molecules.

When we try to discover drugs, we are aiming to perturb disease phenotype to have a positive effect on disease progression. However, we must achieve that by intervening at an individual molecule level, since drugs generally bind to one or a small number of proteins. Thus, for successful drug discovery, we need to understand the basis of this complex genotype-phenotype relationship. Networks of interacting molecules within the cell act together to give rise to function and can be viewed as a mechanistic bridge between the molecular and the phenotypic levels.

We’ve developed a drug discovery process (NDD), and platform that implements that process, around this concept of utilizing sub-cellular networks as molecular correlates of phenotypic features and looking for compounds that can significantly perturb those networks, and by extension disease phenotype.

What are the benefits of this approach?

Our approach is based around thinking of biology from a systems point of view and how that affects the drug discovery process. The use of phenotypic assays, as opposed to isolated protein assays, early in the discovery process is one experimental consequence of this point of view; the use of intact biological systems during screening should result in better translation into more complex systems, ultimately man.

However, there are issues with the use of phenotypic screening. One challenge is that the more complex assays, while increasing translatability, may not be amenable to high throughput screening. They’re harder to run, take longer, and you can’t screen large decks of compounds through them.

Another hurdle with phenotypic screening is that you can identify interesting active compounds, but you have no idea of their mechanism of action.

The NDD approach addresses both issues. One way to view what we’re doing is as an in-silico phenotypic pre-screen. We screen a large multimillion deck of compounds in-silico using our approach, where our network models and analytics act as in-silico surrogates of an assay; and the output is a ranked list of compounds predicted to have activity in a real-world phenotypic assay that represents the modelled biology. That allows us to be much more efficient than a traditional screening approach because the in-silico process can run much more quickly than a real screen.

But, maybe more interestingly, it allows the utilization of more complex real world assays subsequently for two reasons – the in silico process boosts the real-world hit rate by 1 to 2 orders of magnitude (to around 3-10% typically) and allows us to screen 2 to 3 orders of magnitude fewer compounds (based on their predicted ranking) in the lab (typically on the order of 300 to 1,000); the infeasible becomes feasible. This is one area where we’re seeing traction with commercial partners; companies developing complex phenotypic assays for reasons of increased translatability and wishing to use the technology in a more comprehensive screening mode.

There is another benefit over more traditional ‘blind’ phenotypic screening. Because the network models that we build and analyse to drive the selection of compounds are based on real, physical cellular networks, the process gives a mechanistic insight into how the discovered hits are working. This insight can be used to drive mechanism of action studies when further developing identified compounds; and in some cases, can permit early target deconvolution.

An additional advantage of the NDD approach and network biology concepts is that they can support the interpretation of molecular profiling data, such as large-scale transcriptomics or GWAS ata. The interpretation of those data in a network context can be used to identify novel mechanisms involved in the disease processes that can then be used to drive a drug discovery processes, using NDD or a more traditional approach.

Can you tell me a bit more about your partnership with Norvo Nordisk and Type 2 diabetes?

Novo set up an advanced research group in Oxford specifically looking for novel technologies to drive drug discovery. One of the aspects they’re looking at is the use of more complex assays using human tissue-derived assays. But, as discussed above, these assays are generally low-throughput and hard to use in a screening context. Our collaboration is using NDD to come up with smaller decks of compounds that they can then screen in these advanced low-throughput assays that they’re developing in Oxford.

Is there anything in particular that you’re looking forward to?

We’ve spent the last 7 years thinking up the approach, building the computational platform, and validating via our own internal discovery projects. To me, the interesting, exciting thing is going out and talking to potential partners and presenting at conferences, where we’re finding there’s a lot of interest in the approach.

As we’re a small company, we can’t really use the technology and platform capabilities to their fullest extent due to resource constraints. The approach is almost too productive. Working with partners who can enable us to utilize the platform to a much greater extent, to me, as the builder of the technology, is what I’m looking forward to.


Jonny Wray, Head, Discovery Informatics, e-Therapeutics will be speaking on network driven drug discovery in more detail at the Global Pharma R&D Informatics & AI Congress.


To see who else is speaking and for details of roundtable discussions, download the agenda today.

Leave a Reply

Subscribe to Our Newsletter

Get free reports and resources from our world class speakers.
  • This field is for validation purposes and should be left unchanged.