UPPER MERION, Pa.--At the world's biggest corporate bioinformatics department here, the sounds of high-performance servers processing data and scientists producing algorithms never cease. In his new corner office overlooking the sprawling research and development campus here, David Searls, vice-president and director of bioinformatics at SmithKline Beecham Pharmaceuticals, is aware of the constant hum. The day of BioInform's visit to SmithKline's suburban Philadelphia campus, cubicles were in disarray and boxes stacked for an office remodeling project. Having outgrown its old quarters, the bioinformatics group had forced another division off the first floor. Said Searls, "No other part of the company has grown as fast." The same can be said of any other industrial bioinformatics department.
"SmithKline Beecham is the only pharmaceutical company that's made a commitment on this scale," Searls claimed. "Only Glaxo Wellcome comes close."
In 1993, SmithKline's bioinformatics group was set up as a two-person team to handle a database of human express sequence tags being provided to the company through a $125 million deal with the Rockville, Md., biotech firm Human Genome Sciences. Now that 10-year contract and other proprietary and public sources yield a daily flood of genomic data to be downloaded, assembled, and characterized by SmithKline's bioinformatics staff.
In 1995, Searls, a computational linguist with a PhD in biology and a master's degree in computer science, was recruited from an academic position at the University of Pennsylvania to head a department that has grown into a staff of some 70 biologists, computer scientists, and mathematicians in offices at three sites. In Upper Merion, SmithKline's "bioinformatics central," staff is leveling off now, Searls related. Staff here is dedicated to developing software and databases, inventing new algorithms, and support. Twenty minutes away, in Upper Providence, Pa., another group of four scientists, soon to be boosted to 10, is dedicated to microbial research for antibiotics development. In Harlow, UK, north of London, another small SmithKline bioinformatics team works under the direction of Chris Rawlings, who was recruited from the Imperial Cancer Research Fund.
Full Intellectual Partner
Searls attributed the growth of SmithKline's bioinformatics department to the deal with Human Genome Sciences, which "forced us into the genomics age," he said, and to the commitment of upper management, lead by George Poste, the company's chief science and technology officer. "They saw the value of bioinformatics," Searls explained. "In other companies the bioinformatics department is seen as a service organization. Here it's viewed as a full intellectual partner in the discovery of genes and as a contributor to the discovery of drug targets."
David U'Prichard, recently named head of worldwide research and development, has reconfirmed the pharmaceutical company's commitment to bioinformatics. "We're in the process of proving our value by starting to permeate the rest of discovery. The computational assist is more and more important," Searls emphasized. Indeed, SmithKline's bioinformatics infrastructure is an asset that would be difficult to overvalue. Still, the proof of that value in terms of drug development is years away. Dozens of drug targets derived from express sequence tags using bioinformatics tools are in the very early pipeline at SmithKline, but the time to market is more like seven to 10 years, according to corporate spokesman Mike Houldin.
SmithKline's bioinformatics priorities are fourfold, according to Searls. Of primary importance is the development of techniques for identifying and characterizing genes in raw sequence data. Comparing new sequences to existing or known genes is key: a similarity search is conducted to obtain clues about what a gene does. Such characterization is done with a view toward drug development.
The intelligent integration of information is another top priority. Searls's staff is developing techniques for data mining--technology that allows for the discovery of patterns in large databases. Genotyping, or being able to recognize polymorphisms from one human to another to determine drug effectiveness in a population, is another area of concentration. Finally, development of algorithms that will allow more information to be squeezed out of available data is the fourth strategy. For a large number of genes that have been discovered, researchers have no idea as to their function. Developing better algorithms will allow scientists to understand genes in a context.
"In the future bioinformatics will be more concerned with genes and gene products in context," Searls predicted. Indeed, the future is on his mind: 1993 was "the early days," when SmithKline's bioinformatics mission was "overwhelmingly identifying new genes for drug targets," he recalled. Now the objective is shifting to "doing a better job of classifying and subclassifying genes."
"So far, we've been focused on patterns," Searls related. "We will need to understand genetic networks and move beyond simple lists of genes to discovering what gene products do in a context." He added, "There's more emphasis now in SmithKline and the rest of the world on postgenomics. We're anticipating what we're going to do when we have all the genes."
SmithKline Beecham University
As the field advances and competition to recruit from the small pool of bioinformatics talent heats up, Searls anticipates meeting training and education needs in-house. Being the first in industry to spearhead a bioinformatics department has given SmithKline an advantage, he said: with several of the field's pioneers on staff, as well as researchers from a variety of disciplines, the company has a virtual university on its campus here. "We had a clear field to hire for several years and were able to build a critical mass," he commented.
Getting new talent is tougher today, but their top-notch staff gives SmithKline a hiring edge. "The greatest reward is having this concentration of talent to learn from. There are not many careers where you can have this many highly intelligent people from different disciplines on one hallway," Searls observed. "We do very specialized and advanced kinds of training in-house. There's an opportunity to get ambitious about teaching formally inside, with the concentration of people who've been around the field."
SmithKline also maintains close ties to the academic world, especially with the nearby University of Pennsylvania, where Searls worked for four years before making the move to SmithKline. The company supports research contracts and collaborations, and funds some postdoctoral work with outright grants. "At any given time a dozen such interactions are going on," Searls estimated.
Asked whether he has regrets about making the leap from academia to industry, Searls said that despite being "up to the neck" in management, he has none. The SmithKline bioinformatics group has published more than 40 articles this year, four of which he authored. "Increasingly, industry is becoming the best place to do bioinformatics," Searls asserted. "And within industry, this is the best place to do it."
--Adrienne J. Burke