The virtual physiological human: The search for computing’s supermodel
01 June 2009
Is this the ultimate challenge: to use computational techniques to construct a model of how the human body works? Nuala Moran takes a look.
Much of the costly failure of drugs in development comes not because a product is ineffective in treating the disease, but because toxic side effects make it unsafe to use.
But what if rather than waiting until the lengthy and expensive process of preclinical development is complete, it was possible to test not only for side effects, but also activity on the target, in a computer model?
Using computer models and simulations for early stage testing in drug discovery is but one example of the potential applications of the Virtual Physiological Human Initiative, a €72 million European Commission-funded group of projects that are setting down the framework for a computer model of the human body, encompassing physiology from single biochemical pathways, to the orchestration of these reactions in the cells, tissues and organs that make up the body.
Like an architect's drawing, this framework will provide the plan and the standards to allow researchers in any discipline, anywhere, to collaborate and contribute to the building of the Virtual Physiological Human itself. It will also mean the model is constructed in such a way that it can readily be customised, or rather personalised, by feeding in patient data to make the model relevant to a particular individual.
Denis Noble, Emeritus Professor at Oxford University and a founding father of computational physiology, is credited with initiating modelling of human organs, creating the first model of the heart in 1960. Noble, still a leader in the field, has continued to develop his virtual heart ever since, and has been joined by hundreds of other researchers modelling myriad other aspects of the body.
“I would say it is a realistic ambition to model a whole organism. But it will take time and it needs a lot of effort.”
Yaki Setty, Microsoft Research Cambridge
One such is Yaki Setty, of the Computational Biology Group at Microsoft Research Cambridge, who is working on models that simulate the four-dimensional formation of organs. With colleagues, he has developed a model of the formation of the pancreas in the mouse embryo, which traces the formation of the three dimensional structure over time. He is now extrapolating this model to the formation of other organs (see “Modelling in four dimensions”, page 31).
“From the work to date, I would say it is a realistic ambition to model a whole organism. But it will take time and it needs a lot of effort,” says Setty.
Putting it all together
In parallel with the rise of computer modelling of human biology, the reductionist techniques of molecular biology have provided deep insights into the thousands of individual components – genes, proteins, cells, hormones, enzymes and their associated effects on each other and on the body – that are involved in the functioning of the human body.
Now the Virtual Physiological Human initiative aims to make it possible to take these Lego bricks of molecular biology and the partial models developed by Noble and his counterparts, understand where the individual pieces go, and glue them together in a single, comprehensive computer model that replicates the interactions and complexity of the human body.
According to the roadmap for the project, drawn up under STEP (Strategy for the EuroPhysiome, a project funded as part of the European Union's Framework 6 research programme), the Virtual Physiological Human Initiative is not aiming to create the model itself. Instead, it aims to create the structure that will enable researchers working on elements of the overall model to “share observations, to derive predictive hypotheses from them and to integrate them into a consistently improving understanding of human physiology/ pathology, by regarding it as a single system”.
“The ambition and scope of completing the model itself is decades long,” says Peter Coveney, Professor of Physical Chemistry and Director of the Centre for Computational Science at University College London, who is one of the project's leaders. “But the framework we will set up will allow other research programmes and funding agencies to contribute and continue with the work.”
Not only is the framework meant to overcome the division of human biology into organ, tissue, cell, protein and gene, or subsystems such as the gastrointestinal, cardiovascular or musculoskeletal, but also to overcome the fragmentation of expertise into different disciplines, each with a different lexicon. Within the framework of the Virtual Physiological Human initiative, specialists from different disciplines will be able to work together, using a common language.
Building a Virtual Physiological Human depends on recent increases in computer processing power, increasingly sophisticated algorithms and the development of grid computing, that make the project – if long-term – a realistic objective. It also requires a shift in perspective, from reductionist molecular biology, to viewing the human body as a highly organised and massively parallel processing system.
Microsoft Research at the University of Trento Center for Computational and Systems Biology in Italy is working at this intersection of biology and computing to develop new conceptual and computational tools that will enhance understanding of the processes that are responsible for the large-scale properties and dynamics of the human body. One discipline is now feeding off the other, as increased understanding of how the body processes information leads to the development of new, more powerful computing tools, which can, in turn, be applied to biology.
“Increased understanding of how the body processes information is leading to new, more powerful computing tools.”
Into clinical practice
To highlight its practical value and demonstrate that the overall framework is appropriate, the Virtual Physiological Human Initiative is funding a number of individual projects, to show modelling and simulation in action. “The point is that models need to be timely,” says Coveney. “There needs to be the ability to pull out data on an individual patient and run the model fast enough to impact on clinical decision making.”
Rafael Sebastian and colleagues at the Barcelona Centre for Computational Imaging and Simulation Technologies in Biomedicine are building a system, which when fed imaging data from computer tomography, magnetic resonance imaging and three-dimensional ultrasound will create a model of an individual's heart.
These structural models are driven by software that simulates the electrical activity of the heart, making it possible to predict the response to cardiac resynchronisation therapy in particular patients. “Our ultimate goal is to impact on the process of patient selection and the optimisation procedure [that is programming the device] carried out in patients that undergo this complex, expensive and not fully understood treatment,” says Sebastian.
The researchers are currently validating their model against actual clinical data. Sebastian believes that such translation of computer models and simulation technologies from fundamental research into clinical use “is one of the most important challenges in the coming years”.
As the roadmap for the Virtual Physiological Human says, “Currently, we are investigating the human body by pretending that it is a jigsaw puzzle made up of a trillion pieces. We are trying to understand the whole picture by looking only at a single piece or, maybe, a few closely interconnected pieces; it is no surprise that we are not finding it easy.”
When it is complete the framework will enable investigations of the human body as a single – though hugely complex – system.