advertisement
Print

Gosling on Bioinformatics

by Bruce Stewart
02/05/2003

James Gosling is a VP and Fellow at Sun Microsystems. His early years with the company were spent as lead engineer of the NeWS window system. He also did the original design of the Java programming language and implemented its original compiler and virtual machine. He has built satellite data-acquisition systems, a multiprocessor version of Unix, several compilers, mail systems, and window managers, as well as the text editor Emacs for Unix systems.

James is presenting a keynote this week at O'Reilly's Bioinformatics Technology Conference in San Diego, California on Java for Numerical Computing. In this interview, Bruce Stewart talks with James about how Java is being used in bioinformatics, the BioJava project, and the challenges facing bioinformaticists.

Bruce Stewart: How long have you been involved in bioinformatics, and what about it interests you?

James Gosling: To me, bioinformatics has an inherent "cool factor" because it challenges our approaches to analytics, pushes the envelope on performance, and offers a compelling example for a web services-based industry. Not to mention that it might one day have a direct impact on my personal health or the health of those around me.

Sun Microsystems has been a technology provider to the life sciences industry for over a decade. We partner with the industry's foremost hardware and software vendors, industrial companies, systems integrators, standards bodies, academic organizations, and, of course, customers, in developing the solutions that power information technology (IT) across the life sciences industry.

Stewart: What is Sun doing related to bioinformatics?

Gosling: Sun is a prominent player in this area, and we support the software development and go-to-market efforts of nearly all of the leading ISVs in bioinformatics, as well as other focal areas of the discovery and development value chain, which deploy solutions on our technology.

Additionally, through our Center of Excellence (COE) program, we work with some of the world's renowned academic and industrial partners to create communities and forums for exchanging ideas. Sun has numerous Bioinformatics COEs, including the first Center for Bioinformatics in Taiwan; the NRC (National Research Council) of Canada, spanning 4.5 timezones; the VBI (Virginia Bioinformatics Institute), which was Sun's first COE; the University of Cincinnati's Childrens' Hospital; and the University of Delaware. We also have COEs in related areas, such as the COE for Visual Genomics at the University of Calgary (the first Java3D-enabled Sun graphics-based immersive cave).

Finally, our AEG (Academic Equipment Grant) program has provided Sun hardware and support to hundreds of labs working on life science applications, including bioinformatics. Through that program, we have sponsored numerous events, workshops, seminar series, standards, and student-travel stipends to foster the exchange of technology and growth in this industry.

Stewart: How do you see the future of informatics shaping up, and where does Sun want to be in that future?

Gosling: Informatics is evolving away from a "scientific discipline" and more towards a "tool" -- similar to the path molecular biology took about 15 years ago. For a tool to be effective, as well as widely adopted (which is the overall goal of informatics as we see it), it must be truly scalable, intuitive to use, secure, and embedded into a model, whether it be an ontology or other approach to knowledge management, so that using the tool translates to value. Sun believes that tools should be regarded as services, web services in particular, so that they can be used by anyone, anytime and anywhere on the network.

Sun's vision is to be the leading technology partner for the life sciences industry -- enabling our partners and customers to develop and implement best-of-breed solutions, from speeding drug discovery to reducing cost and streamlining corporate functions.

Stewart: Java is in use in bioinformatics. What about it makes it a good fit for this technology field?

Gosling: Data visualization is becoming a critical step in the bioinformatics value chain, and Java/Java3D technology is core to visualization solutions. Researchers want to view three-dimensional models of systems such as genomes, cells, or organs with virtual-reality devices that allow them to see spatial relationships that would be difficult to detect in conventional images and impossible in data tables. A great example is the use of a Java3D-enabled CAVE Automatic Visualization Environment to advance genomics at the University of Calgary.

Stewart: How does BioJava fit into the bioinformatics language landscape?

Gosling: Open source efforts are probably the best way to test out new concepts and weed out concepts that don't work. The BioJava effort, and the OpenBio effort, which Sun supports, among others, all work to reduce redundancy in the workplace and have a direct impact on efficiency. How many BLAST compilers do we need, right? Users can leverage scripts, or whole languages like Java, to ask questions in novel ways that will lead to answers.

Stewart: What do you think are the biggest challenges facing those working in bioinformatics?

Gosling: Everyone is accutely aware of the data tsunami out there, which, compounded by the complexity of multiple data types and a lack of standards and models to interpret all of the data, together comprise the main challenges for the industry -- managing, storing, and analyzing all of this data. As the needs of bioinformatics continue to push the boundaries of IT, the scalability of IT infrastructure, particularly data centers, becomes increasingly important.

I think bioinformaticists today are challenged to use the tools that they already have to answer questions related to a stated hypothesis, rather than trying to use bioinformatics as a science to create answers to questions that haven't been formulated. Again, similarly to molecular biology, it's about evolving from a scientific discipline into an enormously valuable tool and returning to hypothesis-driven research rather than data-driven science.

Stewart: Do you feel that bioinformatics has affected other areas of computing besides life sciences?

Gosling: Because Sun is a technology provider to multiple vertical industries, we're in a unique position to compare and contrast those verticals and how they leverage IT. On the one hand, we've seen how concepts like interval arithmetic have broad utility; but, for whatever reason, industries seem to want to discover solutions on their own. For example, EDA and the oil and energy verticals have been dealing with teraflop clusters for years, but this is a relatively new issue for life science. The same holds true for the film, graphics, and entertainment industries, who have been working with visualization issues for decades but, again, life science as an industry hasn't really leveraged the gains made here either, at least to my knowledge.

It would be great to bring these disciplines together to collaborate on best practices and learn from one another's experiences.

Stewart: Have there been any bioinformatics breakthroughs that have trickled down to the consumer level?

Gosling: There are several interesting breakthroughs on a consumer level that were successful due to the advances in bioinformatics: metabolically-enriched "yellow rice," diagnostics for hereditary breast cancer, variations in leukemia and the first complex human disease gene, diabetes (identified through Dr. Cox's work at the University of Chicago), and the first biotechnology drug, Epogen, from Amgen, for example. Add to that a whole bunch of industrial enzymes that do everything from make our detergents better to digest oil spills ... these are just a few examples!

Stewart: What predictions do you have for how advances in bioinformatics will change our society?

Gosling: I believe that bioinformatics has already changed our society, and, broadly defined, has touched almost every aspect of modern life science research. Quite simply, there is so much data that it has to be processed by computers; therefore, bioinformatics will be at the core of virtually every discovery in this industry -- if it's not already. Over the last ten years, we've seen the emergence of a whole new industry, the employment of hundreds of thousands of people worldwide, new approaches to discovery, new academic programmes to drive creative thinking, hundreds of therapeutics in pipelines, and an insatiable demand for better technology that drives us all to work harder. The next decade promises just as much, if not more!


Return to bio.oreilly.com