The Trento algorithm: A research recipe from Italy

Take an ancient town, a new university, Microsoft Research and a fresh approach to computational systems biology. Mix well. Then apply…

If you drive up from Lake Garda through the vineyards either side of the Adige river you’ll soon come to the ancient city of Trento. Wine apart, it is mostly known for being the town where the Christian world finalised the division between the Catholic Church and the new Protestant ideas in the middle of the 16th century. Yet its university is very young – it dates back only to 1962, and for its first decade it was a liberal arts college with no mathematics faculty. Biology found its way into its curriculum only a couple of years ago.

So Trento might seem an unlikely setting for the first centre that Microsoft has set up jointly with a university, and one that is setting out to transform the way that biology is done, through creating a new partnership with computing.

Yet despite its youth, the university’s science faculty was last year placed on top of Italy’s unofficial university rankings, produced by the Censis foundation. Equally, the university has rapidly accumulated a first-class reputation through the marriage of biology and information science.

Established in 2005 with support from the Province of Trento and the Italian government as well as Microsoft Research, the Microsoft Research-University of Trento Centre for Computational and Systems Biology (CoSBi) is rewriting – literally – the role that computing and algorithms can play in throwing light on nature’s most complicated processes: the workings of the human body.

CoSBi began with just two researchers. Now it has nearly 30, and they come from different disciplines such as computer science, engineering, biology, physics and chemistry. “Nowhere in Europe has so many people focused on algorithmic systems biology,” says CEO Corrado Priami. And it has plans for expansion.

By 2011, the centre aims to have 40 researchers, rivalling any institution in the world for the size and scope of its work in this new field.

‘Nowhere in Europe has so many people focused on algorithmic systems biology.’ – CoSBi CEO Corrado Priami

CoSBi is unlikely to have problems finding people wanting to fill the new posts. In its four years this centre has attracted more than 2,600 job applicants drawn by the Trento approach to computational systems biology.

“One of the exceptional accomplishments of CoSBi is its ability to attract great talents from all over the world, in contrast to the traditional brain drain to which Italy and Europe are historically exposed,” says Davide Bassi, Rector of the University of Trento.

Computing at the core

For Priami, the future of computational approaches to biology is not merely as a handmaiden, providing tools that enable biologists to make discoveries. “Computing and biology have been converging ever more closely for the past two decades, but with a vision of computing as a resource for biology,” he has written.

That convergence has produced outstanding results, such as the sequencing of the human genome. But Priami is looking for something different, something more: a convergence of computational science and biology as a true partnership that will serve both disciplines and which will see computing “not as a service technology, but as an enabling core technology”.

At the heart of Priami’s vision is the concept of algorithmic computation, expressed in radically new programming languages. “Computer science tried to apply its techniques to biological sciences,” he explains. “Good, but not good enough. What we are doing is designing programming languages inspired by biology.”

The idea is to solve one of science’s greatest challenges: how to design new drugs more quickly and more safely. Scientists around the world are looking at simulation as a way of doing just this, as well as avoiding the costly and sometimes problematical use of animals. The use of animals is not only troubling from an ethical point of view: since their bodies are different from ours, drugs that they can tolerate well can prove dangerous when given to people. No wonder policymakers, too, are looking with interest at what computation could provide.

The centre’s strategy is based around the development of a software platform they call the CoSBi Lab. It is designed to be able to replicate, in a computer, all the activities that scientists currently perform in a “wet” laboratory. CoSBi Lab is based on the BlenX language (see Box, “Showdown at Schloss Dagstuhl”, p18), and comprises a suite of five free downloadable software prototypes designed to describe, simulate and analyse different biological processes.

CoSBi CEO Corrado Priami

Paradox of power

Modern computing power and high-throughput technologies are delivering mountains of data. But what biology needs is some way of extracting the general principles that underlie all these data – some way of organising the

information, and then some way of transforming data into knowledge. That’s what algorithms can deliver.

And that, says Priami, entails new programming languages that work the way biological processes do. That’s quite a task, involving as it does languages that accommodate mind-boggling complexity, that solve problems concurrently, and perhaps even solve their own problems.

‘Algorithms highlight the why and the how.’

When it comes to modelling what goes on in the body, raw computing power is never going to be enough: the number of potential interactions between proteins, for example, is greater than the number of atoms in the Universe. That’s where the new programming language BlenX comes in. Developed at Trento, it uses smart algorithms and concurrent processing that prevent the number of variables exploding beyond the ability of computers to handle them.

Corrado Priami puts it like this: traditional computer simulation is essentially about solving a series of equations. The process may tell you what you have got at any given stage, but it does not tell you why you are right, or the relations between the steps you have taken. It does not tell the whole story.

That’s what algorithms can do, because they describe an evolving process. Put simply, it’s the difference between looking at a series of pictures described by equations to looking at a film described by algorithms. “Equations describe the changing of variables’ values when a system moves from one state to another, while algorithms highlight why and how that system transition occurs,” he writes.

Implications

Understanding the why and the how of the way biological systems work opens up the possibility of a better understanding of biological systems that will lead to new, targeted medicines. But it goes further. Systems biology also promises insights into how food can interact with DNA to activate genes that prevent the onset of diseases. And the study of webs of interaction enables the modelling and analysis of ecosystems to determine how the food chain is influenced by environmental change.

“The work conducted at the centre demonstrates the true potential of interdisciplinary research,” says Luca Cardelli, principal researcher at Microsoft Research Cambridge and a member of CoSBi’s board. “This is not just to transfer knowledge or techniques from one area to another, but to use the underlying culture of two disciplines to come up with new approaches and solutions that neither discipline would have imagined. This cannot be achieved by merging research groups: People have to grow into it.”

When Nobel laureate Sydney Brenner visited CoSBi, he left a wry message in the visitor book. “Although I do not really approve of systems biology,” he wrote, “I am happy to be here and to find so many people intent on solving the most fundamental problems of biology.”

Not all biologists are fans of the new systems biology. Schooled in the virtues of patient observation, experiment and intuition, they don’t take kindly to the huge role that now is taken by software and hardware. But they know how important the new field is.

Showdown at Schloss Dagstuhl

Three years after Microsoft Chairman Bill Gates signed the agreement setting up CoSBi, the Trento approach was put to the test in a battle of computational biologists in a German castle. The occasion: a meeting at the elite Leibniz Centre for Informatics at Schloss Dagstuhl in Saarland, discussing the modelling of molecular biology.

With more than 100 researchers from Europe, the US, Singapore and Japan splitting up into teams to see who could most impress the judges – and their fellow researchers – the team from CoSBi picked up the first prize with the CoSBi Lab platform, a suite of modelling and simulation software.

The key to winning the Dagstuhl competition, says Corrado Priami, was BlenX, a language CoSBi has developed to model the composition of proteins and protein domains, and their interactions.

BlenX retains some of the basic principles of stochastic pi-calculus and beta binders – techniques to describe changing systems – but enhances them by taking inspiration directly from the way life works.

What the team from CoSBi showed was that existing models of how cells reproduce themselves (the cell cycle) and tell night from day (circadian rhythms) don’t match what we see in real-life experiments – and they then suggested ways of improving the models. That kind of knowledge has big implications, for example in indicating when might be the best time of day to take particular medicines.

“This external recognition for the work of the CoSBi is a true milestone for the centre,” said Andrew Herbert, managing director of Microsoft Research Cambridge, “and a step further toward fulfilling the founding vision to create a new breed of scientist equipped with the right tools to better understand biological science and help to tackle the world’s environmental problems.”

More information

www.cosbi.eu

Author: Peter Wrobel