Making Science Make Sense
| Home > Scientific Writing and Editing > Feature Article Sample | Home | About | Samples | Contact |
|
Spotfire Turns Up the Heat on Discovery Data
Technology for technology’s sake is easy. The harder job is creating truly useful tools that match, even mirror, tasks. What’s a good example of this? “U.S. gas pumps,” says Christopher Ahlberg, the founder and chief executive officer of Spotfire, Inc., a bioinformatics company headquartered in Cambridge, MA. “To start the pump, you move a handle that blocks your ability to replace the nozzle until the pump is turned off,” he explains. “It’s a small thing, but it makes sure that you follow the right track on the way to accomplishing the task.” Many scientists would argue that Ahlberg’s Spotfire software is an even better example of a tool that matches the task. And investors might agree. In just two years, the privately held start-up has signed on 23 of the top 25 pharmaceutical companies as customers, along with a slew of major biotechs, second-tier pharmaceutical companies, and major chemical companies. The reason for the interest is simple: Spotfire represents a complete departure from traditional approaches to software design in general and discovery data analysis in particular. The Eyes Have It “Oh man!” That’s the response from Ahlberg, a 30-year old Swedish native, when I ask him to rank the usability of the software currently running most of world’s PCs. Certainly graphical user interfaces are a step up from DOS prompts. But now, instead of learning a foreign command language, “we have menus to wade through,” Ahlberg says. “We’re still not interacting with the information itself.” Ahlberg credits “Space Invaders”--the late ’70s video game that is now an icon of pop culture--with offering a better, more intuitive way of interacting with software. “Imagine trying to play this game with a command language: ‘Go left, go left, go right,’” Ahlberg says. “It would never work. Instead, you move a joystick. And you never get a syntax error playing; you never pull the little handle and get a message saying, ‘You moved too far to the left.’ The game just doesn’t let you go too far to the left.” Video games, according to Rock Steven Gnatovich, Spotfire’s president and chief operating officer, are now the paradigm for user interfaces. Software that isn’t “fun” is more likely to be rejected by users. And so instead of overloading software with new functionality, Gnatovich recommends rechanneling technology to provide new ways to look at data. “You can see faster than you can think and faster than you can hear,” Gnatovich points out. For the first time in history, computers have the power to match and even amplify human perceptual abilities. The result is the ability to do “dynamic queries”--Ahlberg’s term, which he coined as a doctoral student of human-computer interaction. The central tenets of this approach are rapid responses (less than 150 milliseconds) to all operations and a highly visual interface where both the objects of interest and the actions are always available on a single screen. Ahlberg specializes in sliders, grabbable bars that, when moved, display a different subset of data. “I’ll probably have a slider on my gravestone,” he quips. His grad school prototypes ranged from a film database to a periodic table application that let scientists move sliders to cluster elements according to attributes like atomic radius and ionization energy. But it was the power of these prototypes to handle large, multidimensional datasets that prompted Ahlberg to tackle discovery information with Spotfire. Looking for Needles by Making More Haystacks Gnatovich, who brought a background in manufacturing IT to Spotfire, notes that on-screen visualization made its way into CAD software when the models became too unwieldy to yield real information. Similarly, the advent of high-throughput screening, combinatorial chemistry, array-based genetic analysis, and robust laboratory automation has discovery scientists producing data like never before--and most existing software can’t keep up. “As a colleague in another software company described it, the data is everywhere and nowhere,” says Gnatovich. That’s because the data is scattered in different formats across distributed databases; the information is there, but unless scientists ask exactly the right questions, they are unlikely to find the specific answers they need. “If you look at this as a ‘needle in a haystack’ problem, pharma’s answer so far to finding more needles has been to make more haystacks, such as genomics, HTS, and combichem,” Gnatovich explains. “Spotfire, on the other hand, is really good at finding needles.” Spotfire’s graphical display offers shapes, colors, and other visual signals to distinguish different data points or subsets. Scientists interact directly with the data, pulling sliders representing different variables to “query” the data rather than entering searches into dialog boxes and clicking “Submit” buttons. And there’s no waiting--tweak a slider, and the display immediately changes, revealing not just specific answers, but trends, gaps, and anomalies in a data set that might have been invisible using a traditional form-based software interface. “Spotfire’s approach has turned a complex set of relationships into things that the human eye can grasp quickly and easily,” says John Sowatski, senior director of software development at Affymetrix. Affymetrix is one of several scientific content providers that has joined the Spotfire Network Partnership Program to link various types of data, including chemical structure and genomic databases, to Spotfire. “If I were working with similar datasets in Excel, it would take me a minute or two to redraw each graph. With Spotfire, it’s instantaneous. It’s not just a great discovery tool, but a good example of software engineering.” Can We Leverage Sweden? Making a different kind of software requires a different kind of company, and Spotfire certainly fits the bill. Board members split their time between the company’s corporate headquarters outside Boston and its development facilities in Göteborg, Sweden. Internationally distributed start-ups are so unusual that the Harvard Business School recently used Spotfire as a case study. But the international split sells well with Spotfire’s customers, many of whom also maintain facilities in the northeast U.S. and in Europe. “We had to be international from the get go—being just a Swedish company wouldn’t sell software.” The split also gives Spotfire a quirky corporate culture. Gnatovich explains that when he visited the Swedish office for the first time, he had to hit his internal “reset” button. “Over dinner, I asked the team to tell me about their families, and they looked at me puzzled and said, ‘What do you want to know about our parents for?’” the 46-year old entrepreneur narrates. “I singlehandedly increased the average age of this company by about 10 years! After a few drinks, I learned that the rollerblading record around the circle of cubes in the office was 23.4 seconds, which, for the first time, was faster than the bicycling record around the same set of cubes.” But Gnatovich notes that it’s the energy of this development team that keeps Spotfire innovative. “We’re the hottest thing going in Sweden,” Gnatovich says, a factor that has helped immensely in attracting the most entrepreneurial developers to the company. “I mean, if we were having to build the whole company here, given the recruiting challenges in this tight market…” The implications give Gnatovich pause. But only for a second. Spotfire is hot, period. Deborah J. Ausman, a freelance science writer based in St. Louis, writes about the tools and technologies that support pharmaceutical and chemical discovery and development. |