Dec. 2006 / Jan. 2007 | TRENTO, ITALY — Microsoft is headline news once more, with the simultaneous release of its Vista operating system and Office 2007. But with less fanfare, it is making its most concerted effort yet to forge links with the life sciences community.
Fifteen years ago, Microsoft established Microsoft Research, which now consists of half-a-dozen research centers across North America and Europe, including outposts in the outskirts of Paris and the foot of the Italian Alps (at the University of Trento). Simon Mercer, bioinformatics program manager, says he expects “more centers will be established around the world in the future.”
Leading the effort in Europe is Andrew Herbert, who heads Microsoft Research (MSR) Cambridge, the first European outpost established in 1997. Herbert, an expert in distributed computing, succeeded the late Roger Needham in April 2003. He oversees more than 100 scientists with specialties in programming languages, security, information retrieval, machine learning, computer vision, operating systems and networking.
“The union of cells and Excel” is a question Herbert gets asked all the time. “The fundamental driver for us in thinking about computer science is the opportunity that computing has to play as a very solid foundation in many areas,” Herbert said during a recent meeting on “Converging Sciences” hosted by MSR in Trento, and building on Microsoft’s superb report on the future of computer science (see “Microsoft’s 2020 Vision for Science,” Bio-IT World, May 2006, p. 6).
“[Researchers] want to use computer technology and IT to reduce the time it takes to arrive at a scientific insight... using a variety of techniques, simulation or visualization or straightforward calculation,” said Herbert. “Very often, they want to analyze large bodies of data, some of which perhaps come from the real world and experiments. Those experiments may be automating data collection. That data has to be curated, managed and shared, very often with international collaborations.”
Herbert sees many parallels between the state of computer science today and the world of office automation in the 1970s and ’80s — an opportunity Microsoft seized upon. “How can you use software technology to help people realize their full potential and give them a more fulfilling and productive working environment? We think the same opportunity exists in the scientific endeavor,” said Herbert. In essence, it is a two-pronged strategy: “How can computer technology accelerate the science of today? And how can IT change and define the science of tomorrow?”
Of course, Microsoft would love to find novel applications for its products in the scientific world, helping to advance scientific computing platforms. Herbert sees real opportunities in improving e-notebooks, grid and web services, scientific workflows, and so on. The pharmaceutical industry is one where government regulation is a big part of the process... it’s ripe for computerization,” he said.
As far as creating new fields of science, Herbert says computer science can introduce new branches of science into biology by “developing powerful computing platforms and algorithms that can make scientific investigations and predictions, and doing science in a more autonomous way.” That is much of the focus of the Trento center (see “The CoSBi Show”).
But Herbert is also interested in the flip side: “What can we learn from science that might impact the way we think about building computers themselves?” Work here focuses on new materials, new computational methods, and ways of making large complex systems autonomous.
This convergence of computer and life science has important implications for the training and funding of future scientists, says Herbert: “Scientists need to be computationally literate and excellent in their own discipline of science. That may impact the way that science projects are organized... there may well be a learning, getting-to-understand-the-problem phase as computer science and scientific ideas are blended together.”
The challenge is, where can we apply the kind of thinking that computer scientists do? Herbert quotes a term “computational thinking,” coined by Jeanette Wing, head of Computer Science at Carnegie Mellon University, who says that computational thinking will be a fundamental skill used by everyone by the middle of the 21st century.
Herbert says MSR research is particularly active in systems biology, where Luca Cardelli has been using a particular mathematical representation developed in computer science to understand biological systems at the cellular level. “To do this is necessarily a partnership with industry. We don’t have all the pieces in Microsoft, we can’t do it all, it’s too speculative,” says Herbert.
Herbert admits that it’s hard to quantify the benefits of such pure research — although he is often obliged to. “When I talk to my paymasters at Microsoft, I give two examples,” he says. “The problems that the scientists are bringing us ... they challenge our computer science techniques. And sometimes the benefits come in surprising ways. Some of our tools for systems biology have been using a new programming language, F#, which in turn has improved immensely from being used on real science.
Herbert cites progress in many areas from high-performance computing and database technology, to the creation of a seamless interaction from science lab notebook through to the long-term archive of publications.
The field poses some interesting questions about Microsoft’s attitude to intellectual property emerging from such research. “The research has its own justification, but when we start thinking about the commercial side, the opportunities are very exciting,” says Herbert.
“If we were a pharmaceutical company like Novartis, we’d directly exploit the scientific results. But as a software company, it’s less clear,” said Herbert. “Our researchers publish and share their software tools with the research community but we protect the IP we create so it can be exploited.” Clearly much of this research will yield niche applications — software products and services that don’t fit the current Microsoft businesses. “What do we do with these? Are they something we’d grow as our own business? Would we license the IP to our partners to exploit? Would we launch as a business in its own right? Certainly for pure science results, it’s unlikely Microsoft will go into the drug business or the engineering business, so its more likely an opportunity for joint ventures. As a company we’ve been developing programs such as IPVentures to make this possible.”
Web-Only: Relationship Building in Synthetic Biology
Email Kevin Davies.
Subscribe to Bio-IT World magazine.