Scientists use computer modelling for drug discovery

We now live in a 'Big Data' world and computational modelling of biological data, or computational biology, is seeing new approaches, developments and revelations in the way we discover, deliver and understand the effect of drugs. Although Pfizer has withdrawn its takeover bid for AstraZeneca, the progress that it has made in its cancer immunotherapy research using computational biology was one of the key reasons that Pfizer was interested. It's also a very big part of why AstraZeneca was able to resist the takeover bid.

Computational biology has come on in leaps and bounds in the past few years as supercomputing and biomedical data have combined to great effect to improve our understanding of a range of things from how the body will react to drugs to why natural substances such as green tea are good for us, and also to reveal the links between certain genes and diseases.

Relating and constraining digital cell behaviour

There are two main approaches currently used in computational biology: relational databases and constraint-based models. Dr Nick Plant, reader in molecular toxicology at the University of Surrey, recently presented workshops looking at the theory behind them and how they work in practice at the 9th Annual ADMET 2014 conference, which is a major event in the pharma-industry calendar for those involved in drug discovery.

Dr Plant explained that relational databases do exactly what the name implies: they are used to relate bits of information to each other. "In a biological context, this may be as simple as the interaction of two proteins, or a more complex behaviour such as the general biological response to a chemical exposure."

So, in layman's terms, relational databases are basically our repositories of knowledge. Dr Plant compared their use to that of a dictionary, whereas the constraint-based models build upon the knowledge contained in the relational databases to turn them into 'living' computer models that can replicate biology. "They are called constraint-based as they are constrained by this knowledge, such that they can only simulate things that we have said can happen, and cannot undertake actions that we have forbidden."

This means that researchers use the dictionary contained in the relational database to generate a working model of the cells in our bodies, aiming to reproduce in a computer what we see in real life. Dr Plant says: "The reason for doing this is that in making these constraint-based models we can see new properties emerge; an analogy is that a dictionary is a set of definitions, but a very boring read – put the words together in the correct order, however, and you have a best-selling novel."

Telling biological stories

The creation of these best-selling biological novels is revealing some incredible links between genetic reactions and has also just revealed that the number of genes in the human genome is fewer than we previously thought, and means we have less than an earthworm. Dr Plant explained that as to how super-computing works in terms of biology, perhaps the greatest example of emergent behaviour is life itself. "Looking at our DNA, which can be viewed as the body's blueprint, I cannot predict what a living cell will look like, but using information from a relational database, I can constrain a model in such a way that it really does come to life in the computer and begins to look and behave just like a real cell."

What constraint-based models really do is enable us to begin to understand how different behaviours emerge from the blueprint of our DNA, so-called genotype-phenotype relationships. At Surrey University the research that Dr Plant and his colleagues are involved in is advancing our understanding of two of the biggest health problems we face in the UK.

"We have two great medical challenges: obesity-related illness and cancer," says Dr Plant. "We are still unsure how these diseases emerge from normal physiology, why some people are more susceptible than others, and how best to treat each person's disease, which is also known as personalised medicine. At Surrey, we are building models of the changes that occur within the liver during chronic over-feeding. This will, we hope, help to explain the mechanisms by which obesity leads to disease such as metabolic syndrome, diabetes and liver failure, and how to prevent and treat this."

Genetic revelations through personalised models

In a related project, Dr Plant's team have also looked at data on the genetic make-up from 2,000 breast cancer patients to build personalised models of their tumours. This has allowed them to identify the characteristics that lead to aggressive tumours with a poor patient survival rate. "We are using these characteristics to develop biomarkers that help us to identify these individuals. Then we can develop treatments specifically aimed to be most effective in this currently poorly responsive group of patients."

At present, constraint-based models are largely restricted to simulations of single cells and the short- to medium-term goal of the computational biology community is to create joined up models that allow more complex parts of biology, such as whole organs or tissues, to be modelled. Dr Plant said that there are already exciting advances being made towards this goal: "Using software developed at Surrey, we are currently making such multi-scale models of the liver that truly live and breathe, as they are able to respond to changes in the environment. As we incorporate more features in these models they become more accurate and can reproduce more of the abilities of complex organs such as the liver."

In the long-term, it's hoped that they will be able to join all of these models together to create true digital organisms that are faithful representations of all the complexities of life within a computer. "Such advances will allow us to better predict how the body will respond to, for example, changes in its environment or the development of disease. This will allow us to target our experimental work more effectively, with the twin aim of bringing safe, efficacious drugs to market sooner, and reducing the number of animal experiments that are required, as we will have digital rats, mice, dogs, etc, as well as a digital human."

As well as the revelation that the human genome has fewer genes, and the exciting developments in obesity and cancer research at Surrey University, there has been a flurry of research discovery announcements in the computational biology world this year.

Alzheimer's and brain cancer

At the Houston Methodist Research Institute (HMRI), a team led by Dr Stephen Wong, the founding chairman of HMRI's Department of Systems Medicine and Bioengineering, has revealed a link between Alzheimer's, the most prevalent form of neurodegenerative disease, and glioblastoma multiform (GBM), which is the most aggressive form of brain cancer. 

Using supercomputers at the Texas Advanced Computing Center (TACC) at The University of Texas at Austin, the team analysed and compared data from thousands of genes in order to narrow the search for the common cell signalling pathways of these two diseases. What it revealed is that they share a pathway in gene transcription, a process essential for cell reproduction and growth. What it means for treatment research is now that the pathway has been identified, the information can be used to design a new therapeutic strategy that targets that pathway.

Green tea's role in lowering cholesterol

The health benefits of drinking green tea have long been recognised and it has been a major component of Chinese medicine for centuries. However, research scientists at Sun Yat-Sen University in China have now used computational biology modelling to reveal how green tea metabolites act on human enzymes to lower blood cholesterol levels.

With 2012 statistics showing that six in ten adults in the UK have raised blood cholesterol levels, and the 2004 INTERHEART global case-control study (one of the largest and most important cardiology studies ever conducted) estimating that 45 per cent of heart attacks in Western Europe are due to abnormal blood lipids, this is an exciting discovery that is expected to lead to a real breakthrough in prevention and treatment of cardiovascular disease.

Dr Jose M Prieto-Garcia, lecturer in Pharmacognosy at the Centre for Pharmacognosy and Phytotherapy at University College London (UCL), said that the potential impact of this discovery is that the pharma-industry could now use these products as lead compounds to develop more potent synthetic ones and provide new medicines for the treatment of cholesterolemia, which plays a major role in the development of cardiovascular disease.

"Catechins inhibit many enzymes in vitro, so could potentially exert a positive influence in the metabolism of mammals," says Dr Prieto-Garcia.

These catechins are found in many fruits and vegetables, known as superfoods for their role in keeping our bodies fit and healthy, as well as in widely consumed leafy plants such as tea. Green tea is one of the most accessible and richest sources of catechins, but other important sources are dark chocolate, cherries and red wine.

Prior to this recent discovery, Dr Prieto-Garcia said that because catechins can polymerise into tannins, which precipitate with proteins and lead to the 'dry mouth feel' (astringent effect) of tea or some red wines, many research scientists questioned the in vitro effects of catechins and thought that they were a consequence of unspecific precipitation rather than specific interactions.

The research team in China tested the activity of three enzymes, which are essential for cholesterol biosynthesis in vitro, in the presence of four different polyphenols that are found in green tea. They found that two of the polyphenols could inhibit all three enzymes simultaneously, whereas the other two polyphenols had no effect whatsoever. The team then used computational modelling techniques to work out how the inhibiting polyphenols bind to the enzymes to prevent their function.

Dr Prieto-Garcia says: "Essentially, what this research demonstrates is that catechins can – in theory – interact with the active sites of three enzymes that are essential for cholesterol biosynthesis. So, potentially, a regular intake of catechins will result in a decreased de novo synthesis of cholesterol and help to lower levels of it in the blood."

Computational biology at UCL

The development of computational biology techniques is also benefitting the research that Dr Prieto-Garcia is doing at UCL. This is currently focused in three key areas: herb/drug interactions, the use of natural products against cancer migration and the prediction of bioactivities of complex natural products by artificial intelligence.

"In the past, we could only make relationships between a few parameters and the pharmacological effect. With these modern computational methods we can now make relationships with all the data we have and the observed effect, so we can effectively pinpoint which parameters are more relevant, thus speeding up research."

His research involves looking for functional compounds from plant extracts containing dozens of components, which could be added in small amounts to improve the efficacy of existing drugs and/or herbal remedies to improve performance. It is also focused on discovering the synergies and antagonisms between functional compounds and computational biology is moving research into all three areas forwards. "By applying computational methods we can infer which subset of natural compounds is linked to the pharmacological effect, which allow us to isolate the bioactive fraction faster."

But speeding up the rate of discovery and isolation of these compounds is not the only benefit that computation biology is bringing to Dr Prieto-Garcia's research: "On the other side we can model how the natural molecules actually work on our body.'By doing this we can propose the mechanisms that should be used to get the effects we want, then validate our proposals. All of which is resulting in a better understanding of the therapeutic potential of these natural products."

So what are the healthcare benefits that patients can expect to see from these research advances? They include faster development and approval of new drug treatments. "It helps speed up the 'bench to bedside' process. Namely, more natural promising molecules can be fed into the drug pipeline and eventually one could become a medicine," Dr Prieto-Garcia explains.

However, there is enormous potential for these technologies to help people in advance of needing drug treatments and also for the health benefits of natural compounds to be available much faster than drugs that have to go through years of testing and approvals processes. "When formulations of natural products from plants that are already present in the diet show promising functional properties, they can quickly be provided to the public in the form of food supplements (or 'nutraceuticals')," says Dr Prieto-Garcia. "They may help the population to slow down the onset of some conditions associated with the normal ageing process."

Long-term achievements

The advances that computational biology is bringing to research at UCL, and other universities and pharma companies, is playing a vital role in moving towards achieving the long-term goal of healthcare researchers the world over: to prevent rather than cure diseases.

Dr Prieto-Garcia concludes: "This is especially important in chronic disorders associated with ageing. In my research at UCL, computational biology is helping us to pinpoint more and more the naturally occurring products in our food, which, properly enriched and formulated, may help to delay or mitigate the deleterious effects of such conditions. Once the disease is present, medicines inspired in such natural molecules could provide us with the enhanced pharmacological effect to treat them."

Mental health advances

So as the physical health problems we face and how to treat them are becoming better understood, supercomputing is now being adopted for 'computational psychiatry' to try and make the same level of advances in our understanding of mental health disorders.

The world's first computational psychiatry centre opened in London in April 2014 and it has been funded by a five-year £4.1m investment from the Max Planck Society and UCL. Researchers there will use powerful modern technology to create more detailed models than ever before of how the human brain works.

These models will then be linked to measurements of behaviour and changes in brain function to help identify the causes of a range of common mental health problems so that personalised treatments can be developed.

The Centre will be based between UCL (University College London) and the Max Planck Institute for Human Development in Berlin and led by Professor Ray Dolan FRS, director at the UCL Wellcome Trust Centre for Neuroimaging and Professor Ulman Lindenberger, director at the Max Planck Institute for Human Development.

It's expected that the discoveries made there will spearhead new perspectives on psychiatric disorders that will realise major benefits for patients within a decade.

So if you combine the advances of sculpting drugs to treat physical ailments with those to treat psychiatric ones, and being able to tailor them to individual patients' needs, then it seems that the future for human healthcare that's been created by supercomputing is truly revolutionary.