Executive Roundtable Participants
Andrew Grimshaw, CTO, Avaki
Michael Griffin, CTO, Xpogen
Michael McManus, COO, AnVil
Evan Steeg, CEO, Molecular Mining
BioInform played hooky during the plenary talks during the recent TIGR Genome Sequencing and Analysis Conference in Boston to sit down with four local executives and chat about the state of the life science IT industry. They may have disagreed on a few issues, but they did have one thing in common: None of them wants to be known as a bioinformatics company.
BioInform: Where do you see the genomics industry heading and where do you fit into the picture?
Griffin: We’re in the business of confidence building. Despite the claims in this industry that there’s too much data, there really isn’t enough. By my back-of-the-envelope estimation, I’d say there are orders of magnitude more data that need to be collected in order to make conclusive statements about that data. So in order to build confidence in the data that you have, you have to bring in different sources of data and that’s where our company comes in.
McManus: You have an interesting background. Fluid dynamics, is that correct?
Griffin: Computational fluid dynamics.
McManus: So for you a lot of data is a different story.
Griffin: In this space, it almost seems to me that there’s a deliberate naiveté. Biologists understand that biology is very complicated, but then they go out and collect a little bit of data and expect to get conclusive evidence about some biological function or process… Everyone’s talking about the failure of genomics now, but that’s because it was overblown; it was hyped. You end up dealing with a real biological system and it’s incredibly complex, and yet you collect data from four or five samples of cancer patients and you expect to find some evidence as to what is causing that cancer. I don’t follow that reasoning.
Steeg: We encountered situations within some large pharma organizations where it’s clear that people wanted to put their foot in the water, they wanted to test microarrays or high-throughput proteomics platforms, but they didn’t spend enough to get a real return on it — they spent just enough to get somewhat frustrating datasets.
Griffin: I think what we’re going to be reading about in 12 months is the failure of pharmacogenomics, because people are not collecting enough data in that space.
McManus: Pharmacogenomics has been considered a bad word in the VC community. It’s like bioinformatics. It’s unbelievable. But we all know that medicine doesn’t affect everyone the same way. The question is how do we take the pharma companies and move them through a transition?
Steeg: It seems to me there are three questions that will govern whether pharmacogenomics will happen. One is, is there an overall economic argument for it? I think probably the answer is yes. The second question is, is it technologically and scientifically feasible? I personally believe it is. The third one is, is there a business model to deliver it? And this is what stops the VCs. We all agree it’s going to happen. It’s like systems biology — that’s now a good buzzword, bioinformatics is a bad buzzword — we all agree that at some point in the future people will be doing a lot more biomedicine on the computer, but what is the right business model to deliver it?
McManus: There are two business models here. One is to get a VC’s attention and then there are the pharma companies. They are clearly working on it. They’re not going to figure out the business model that will allow them to make money, but they clearly aren’t just burying their heads in the sand.
Grimshaw: Those are companies that can have longer timelines. If you’re a venture-funded company, your time horizon is typically cliffed at about five years, except in some pretty rare cases.
BioInform: How are you handling that gap between what the VC community expects and the actual timeframe of the pharmaceutical pipeline?
Griffin: Trying desperately to stay away from VCs. We don’t understand where we’re going to be in five years, so how can we pitch it to a VC that’s interested in a $100 million company? It’s not even worth wasting our effort. So we’re driving toward cash-flow-positive as quickly as possible.
McManus: There are other sources of funding. We’re in the services business and a lot of what we’re doing is revenue-share based, so some of the cash reward that we would get for being successful is deferred. But the problem is you need to be capitalized. If you’re lucky enough to get a piece of a drug — we’re not yet, but if you do — you’re pushing that out nine, 10 years, depending on where you are in the process.
Steeg: But if you can do that, that is a great proof statement to take to the VCs if you need to.
McManus: Actually, the thing is, once you do that, you don’t need the VCs.
BioInform: You’ve brought up a number of buzzwords that have come and gone. How do you avoid becoming members of the buzzword graveyard?
McManus: It’s tough. Bioinformatics has a bad name, like genomics does, for the social reason. Genomics went up so high because there was this great hope that it was going to solve all of our medical problems. And it didn’t. I think if you ask the average, non-scientifically aware individual in the country, they thought there’d be this huge change.
Grimshaw: Nobody who understands the scientific process, whether it’s in biology or anything else, would have believed that. That’s always just the first step, collecting the data, and then you’ve got to figure out what it means.
Steeg: I think part of it is just the changing business environment overall. One buzzword that I think will never go out of style is ‘sustainable revenue stream.’
Grimshaw: Actually, it did two, three years ago. [laughter]
BioInform: You’ve identified a need for the services you provide. How do you make that need clear to your potential users?
Grimshaw: We focus on cost savings, because that’s what everyone wants to hear right now.
Griffin: For us it’s all about proving their data. We insist on running trials with customer data and our software to prove you can find things there that you can’t find anywhere else on any other tool.
Steeg: Typically on the collaboration side, you start with a pilot, you show the results, and we’ve done that very successfully in a bunch of different areas. On the product side, I’ve been somewhat surprised by the degree to which we have to do that as well. The bad thing is that that model doesn’t scale. You can’t do demos for every placement you want to make in the world.
Grimshaw: This isn’t unique to this industry or any other. Any time a new technology comes into any area, there’s an adoption cost. So we try to push that adoption barrier as low as possible by making it look as much like what they’re used to doing already as we possibly can.
Griffin: We have a little game we play, the drug discovery game. You create teams of people and they all have thirty minutes to find a drug target and tell us when they find it, and what’s the biology behind it. And it works. It’s fun and people get into the product that way and they enjoy it. It sounds kind of trite, but you do have to make it fun.
Steeg: You have to have some value proposition that you can prove. And the degree to which you can do that quickly depends on what the value proposition is. If it’s time savings, for example, or saving FTEs, that’s more immediate. That’s an easier sell than, ‘Here’s a novel target with more knowledge about mechanism of action than you had before, therefore, if you screen leads against that you’ll have more tox knowledge and blah blah blah.’ Well, how long will it take to validate that?
McManus: Boston Consulting Group estimated that there are 160 companies in this space, and they say that people are bewildered. So I think a lot of what’s going on in the market is that it’s a waiting game that says some of us aren’t going to be here in a year. But I think what they’re failing to realize is, depending on how the market goes, there may be 500!
Steeg: This may not be a good long-term thing, but maybe there’ll be five big behemoth companies that have the whole thing — data integration, knowledge management, data mining, prediction and in silico biology. That will certainly make choice easier in the short term.
Grimshaw: Choices are easier in a Soviet supermarket, too.
McManus: But in a pharma company they’ll wait. They’re doing stuff internally anyway, so they can wait six months for the market to settle down.
Griffin: It’s not one size fits all in this space. In fact, I don’t think there’s any one-size fits all product in any kind of research. If I were a VP of R&D in a big pharma company, I would say, ‘I’ve heard enough claims from all these companies out there. I’m just going to hire a roomful of software engineers, and develop my own tools.’
Grimshaw: When it comes to system software, the days of individual companies being able to run their own systems software passed thirty years ago. It’s incredibly expensive to maintain, let alone build.
McManus: It depends on how much the scientists and technical people in the company think they know about it.
Grimshaw: Actually they’ll say, ‘Look, we can do that with Perl scripts any day.’
McManus: If I hear Perl and Python one more time…but I’m talking about infrastructure stuff. Most scientists I know won’t delve into that area.
Steeg: Some of the scientists believe that they can develop professional-quality software that does all the informatics in bio and chem and so forth, and in some cases yes and in some cases no. But it’s always close enough to their intellectual core competencies and core interests. Everybody can hack Perl and everybody can hack Python, so they’re more tempted to try to duplicate the kinds of solutions that those of us provide at that level.
BioInform: Do you find, as small companies, that potential users aren’t confident you’ll be around in a few years? Or have some of your predecessors who have already disappeared spoiled things for you?
McManus: It’s a real issue, but at the same time innovation is what you see in these small companies. I think innovation happens more rapidly and to a greater extent than it does in a large company. So if someone’s going to be that close-minded about taking a risk, then they’re going to suffer on the innovation front.
Steeg: Whenever a competitor crashes and burns, there are three levels of reaction. The first is, ‘Oh that’s unfortunate,’ the second is, ‘Woohoo!’ The third is, ‘Uh-oh….’
Bioinform: Is the market for your products as promising as some estimates have made it seem?
McManus: If you think of the market as the amount of R&D dollars that are spent in pharma companies that could be allocated to the stuff we do, that’s a sizable market. The question is, how much of that’s available? And if it’s all a not-invented-here issue, then the available market’s much smaller. But pharma has outsourced a lot. The outsourcing of clinical trials is completely routine. And you see outsourcing of research, so the same argument can be extended to using an external party to help develop what they’re doing.
Steeg: If you go to one of these ‘future of the pharmaceutical industry’ conferences, some visionary from one of the Wall Street firms will talk about the future of pharmaceutical companies being very much like some of the automotive companies, where they’re his shell that only does the assembly and the marketing. And then two speakers later there will be someone who will talk about how, actually, no, pharma’s going to grow and consolidate by buying all these different vendors. I haven’t seen strong economic arguments either way.