Six bioinformatics execs shoot the breeze, and ponder the future of their embattled industry
By Bernadette Toner
The bioinformatics sector has hit its share of speedbumps in the past year. Software companies Genomica and DoubleTwist disappeared. Lion, which has hopes to achieve profitability by 2004, said it would cut 80 staff positions along the way. And InforMax saw its stock plummet to an all-time low this spring before warning of gloomy first-quarter earnings.
To take the pulse of other industry players, GenomeWeb recently met with six bioinformatics leaders to chat about the challenges and opportunities they face in the months and years ahead. Over steaming bowls of seafood chowder in a room in Boston’s Exchange Conference Center, participants took in the view of Boston Harbor while engaging in a lively debate over the future of the sector. The participants are: Richard Gill, president and CEO, AnVil; Soheil Shams, president and CEO, BioDiscovery; Doug Bassett, VP and general manager, Rosetta Biosoftware; Dave Snyder, CEO and chairman, Acero; Keith Elliston, chairman, president, and CEO, Viaken Systems; and Sandip Ray, founder, president, CEO, X-Mine.
(A version of this discussion originally appeared in BioInform.)
Market estimates indicate there’s a great deal of opportunity for companies involved in bioinformatics (however you decide to define that term), but there seems to be a disconnect between the financial forecast for these tools and the poor performance of companies in the sector so far. How do you explain this gap and what is your company doing to ensure success in the marketplace?
Gill: At AnVil, we use our proprietary software in in silico drug development, so it’s more broad than bioinformatics or the intersection of IT and biology — it’s about extracting knowledge from biological data. We’re finding that the business is much like it was 10 years ago when pharmaceutical companies were first deciding to outsource clinical trials. It’s a similar psychology right now. The primary issue for the industry is it has to gain the trust of its clients. You have to show you can provide solutions.
Snyder: [Acero has] Incyte’s [Genomics Knowledge Platform] technology, underlying technology from Secant, and a small professional services group. We have Incyte as a channel provider and an investor. GKP addresses the data integration issue, so it’s solving a problem that has gone unanswered for a long time and the initial feedback has been good. I have to say, as a business and technical guy, that I’m very encouraged by this field.
Shams: BioDiscovery started in 1997 so we’re one of the earlier companies in this space. We started developing tools for the gene expression market. In bioinformatics, the problem is that the term is vague. Most people, when they talk about bio-IT, are talking about everything from hardware to infrastructure, to databases and storage. At the time I first looked into this market, everything was about sequence analysis. It wasn’t very exciting in terms of the technology you could apply or the computational methods. But now the complexity and size of the data has gone up, especially with regard to gene expression and proteomics.
There’s a huge influx of data so we’re turning to advanced methods of signal processing and a repertoire of computational methods that exist outside of this domain for solutions. Now we’re seeing the data that can take advantage of these tools. There’s an imbalance of supply and demand — pharmaceutical companies need to crunch this data, but people with experience doing it are in limited supply so there’s a huge opportunity for companies like BioDiscovery with both the experience in complex data analysis as well as the various nuances of microarray technology.
Ray: X-Mine launched in September 2000 with the goal of interpreting multivariate data of any flavor — gene expression, mass spec, IC50, and so forth. I’m a Drosophila geneticist by trade. Todd Golub at the Whitehead Institute and I, along with Rob Tibshirani and Trevor Hastie from Stanford, saw fairly early that off-the-shelf methods for extracting patterns from multivariate data would create a bottleneck. Conceptually, X-Mine wanted to create a systematic approach for data mining and knowledge discovery that would be accessible to biologists at a meaningful level, and we have succeeded in developing such a system.
We are distinguished by a number of supervised learning tools, which allow researchers to pose biologically relevant questions when analyzing large and complex data sets, with emphasis on rigorous statistical validations that are understandable. The user experience is further enriched by a unique information-mining system that is able to search for functional relationships between genes and genes, genes and tissue, genes and drugs, and other relationships from large textual repositories, such as Medline. This is exciting. At a time when drug development is increasing in cost, we are in a unique position to impact major diseases sooner and to impact the cost and time factors directly. [That’s] because we have developed a very creative approach for the interpretation of data and information that can help companies make decisions faster in bringing effective, targeted drugs to market intelligently.
Bassett: Rosetta Biosoftware is focused on commercializing enterprise software for gene expression analysis. There’s a lot of power in synergy and integration, but companies that try to bite off a little more than they can chew end up with a mediocre solution. We don’t want to deliver anything to our customers but the best solution. The downside of that, of course, is that we can’t offer a complete end-to-end solution that allows integration across different platforms and data types. For that reason we’re very involved with several standards organizations to make sure we deliver products that link into our customers’ systems. We also have a professional services organization, but our main focus in on the delivery of shrink-wrapped software.
Elliston: Viaken focuses on informatics from the solutions view — not at the top of the market, but in the mid-tier market. Our infrastructure-based offering is a key technology for smaller customers who are not in a position to hire three people to manage their enterprise bioinformatics system. There’s a dearth of experts in the bioinformatics field who are able to effectively mine data. We help our customers commercialize that data and we’re looking to generate intellectual property of our own in that process.
What do you see as the biggest challenges for bioinformatics companies going forward, and how is your company addressing those?
Elliston: One of the key things I see is that the history of bioinformatics has been focused on the top-tier customers: the Mercks, the Pfizers, the GlaxoSmithKlines. Bioinformatics companies who have less-than-stable valuations right now have serviced only the top percentage of available customers, but these companies already have a strong IT infrastructure, so the level that you have to invest to add value to that customer is very high.
Pharmaceutical companies spend only 35 percent of the total bioinformatics investment; the remaining 65 percent is spent in biotechs. So we’re targeting our offering to the right area of the market, but the challenge is that customers in that market don’t know what bioinformatics is or what it can do. We have to get over a tremendous amount of hype. We have to target the part of the market where we can build a real business and then target the top of the market.
Snyder: We were called into a small biotech firm in Montreal who had spent $4 million in the last 18 months on their own integration platform. They did a comparison and came to the realization that ongoing maintenance on the system would have killed them. We’re selling our system on a subscription basis and it includes maintenance. So you’re right about going after the middle market. It’s amazing to me to see what they’re able to spend, but they’re competing against guys spending big bucks on IT.
Elliston: It’s a question more of an inexperienced market.
Gill: How do you blend domain expertise with the technology being offered? It can be enterprise-wide or solutions-based, but the middle ground is dangerous. A lot of people are generating data right now because they think it’s their job. But if you listen to the mature players in the market, they’re not saying they want more data or they want their data integrated, they’re saying they want their data interpreted and analyzed and made into knowledge. There’s a business there — the real value is how do you get knowledge out of that data. All of us are in that space. We’re trying to provide that bridge and help them take the next step. We’re holding their hands to give them the information they need to make decisions.
Right now, there’s an estimated 160 of us in this space — call it companies who don’t want to be called informatics companies. As an industry, we’re doing the client a disservice right now. There are too many options. There won’t be 160 companies in this space in six months’ time. Out of those 160 companies, how many are announcing deals?
Elliston: A lot of clients don’t like to talk about hiring bioinformatics companies.
Would you say you’re seeing a Celera effect? There was a lot of hype around the mapping of the genome, but having it hasn’t gotten anyone to the next big point yet. Does that say anything about the maturation of the industry?
Elliston: I’d say the concern is more that the market valuation of companies will be below their potential.
Bassett: We as an organization believe very strongly in the power of software to solve the big problems in life science research. We’re not going down that other path where we’ve decided we’re now a pharmaceutical company and will bulk up our chemistry capabilities.
Gill: Some companies can do that now because they’re actually banks…
Elliston: The question is, are you selling your technology to a huge market or are you using your technology to build what the market would build? If your platform technology is unique, then it’s a valid model. Take the example of MSI and Vertex: one sold its technology platform and one used it to build its own products. If you’re using the technology to produce an end product, then that’s a valid way to produce value from that technology. If you look at the R&D investment that goes into this space, software should be a good business because of its gross margins, but you can’t look at this like a typical software business.
Bassett: Analysts liken biotech to the gold rush and call the tools companies the ones selling picks and shovels.
Elliston: But if you take that analogy further, most of the goldminers never found any gold.
Bassett: My core competency isn’t digging for gold.
Gill: You can have your gold mine and also sell the picks and shovels.
Shams: But in this space, the picks and shovels are quite complex.
Gill: I like to think of AnVil as having built a concept car — you are welcome to borrow it and drive it around (and even wash it and service it), but if you want to get past the checkered flag first then the AnVil team should drive it for you.
Elliston: But how do you get a reasonable return on investment for your investors saying you should look more like a software company? I think two paths will come out of bioinformatics — that path, and the software product path.
Bassett: But organizations won’t be able to walk the middle of the road.
Where do each of you see your companies on either of those two paths?
Bassett: We sell software to the marketplace. We are not a discovery organization. We’re out there to empower those organizations.
Gill: We develop solutions and our aim is to support solutions. We give clients access to our people and technology and we undertake the work alongside them.
Elliston: I think it will become black and white, but today it’s not. We’re looking at it more holistically and that’s taking us down the discovery path. But we’re not software developers. MSI/Vertex may be a black and white example, but Millennium started as a platform company and worked further and further down the chain. Now it doesn’t have to partner with anyone. That’s the path Viaken is walking. We’re enabling our customers by generating intellectual property. Eventually, we’ll have enough of an IP base to be standalone.
Gill: You forgot the “but” about Millennium — they’re not a fully integrated pharmaceutical company yet. They’ve in-licensed all their major products. That’s integrating by buying all the pieces.
Elliston: A discovery company and a fully integrated pharma are two different things. Millennium is a good example of the first. I think the pharmaceutical industry will end up something like the auto industry — pharmaceutical companies will put together the components from their biotech partners.
Shams: We provide complete solutions. Initially this meant a fully integrated system that manages and processes the data at every step of an array experiment, and now we see the real need for providing our customers with assistance in using these tools to meet their desired end-results. This means providing both training and collaborative research support.