New entrants to the bioinformatics software industry should heed its history
By Aaron J. Sender
In the penthouse of the Pepper Hamilton law firm in DC one afternoon in December, a group of about 20 attorneys, VCs, laid-off dot-commers, techies, and other bioinformatics business wannabes sat around a conference table. On the agenda for the monthly meeting of the newly founded BioIT Coalition, whose mission is to promote the convergence of bioscience and IT in the DC area, was “Obtaining funding for bioinformatics companies” and “How can IT companies ride the bioinformatics wave?” The banter quickly revealed that this group has a postmodern notion of what the bioinformatics industry is: a field rife with heavy investment and big, fully integrated platforms that will supposedly transform the future of drug discovery.
Their perception isn’t surprising. The industry’s recent history is full of stories about hundred-million-dollar private investments and public offerings, and promises by companies to provide complete informatics systems that offer everything for analysis of DNA chemistry, drug discovery, toxicology, data management, and even diagnostics.
Lion Bioscience, for example, raised more than Euro 200 million in its August 2000 IPO, on CEO Friedrich von Bohlen’s promise that the company would “go beyond software and become an integrated part of the health company,” and “offer a solution for the complete R&D process.” Other bioinformatics software players such as Genomica, Molecular Applications Group, NetGenics, and Oxford Molecular, among others, filled their coffers by making similar pitches.
A History Lesson
To justify this magnitude of investment, one must gamble that a several-hundred-million-dollar market will, within the next few years, be worth tens of billions. Von Bohlen argues that a “[tens of] billions dollar market five or six years from now is not unrealistic.” But if you’ve been around since the early days you know that such bets are a little wild.
Remember Myra Williams? In 1999, VCs ousted her from the CEO position at Molecular Applications Group and divvied the six-year-old company’s assets between Affymetrix and Celera. Oxford Molecular, having given the integrated platform strategy more than a 10-year chance, sold its software business to Pharmacopeia for less than $30 million in 2000 and then disbanded. And Genomica, which in 2000 boasted the biggest US bioinformatics public offering ever — $122 million — became the poster child for the bioinformatics-boom-gone-bust a year later when Exelixis bought it simply for its cash.
Newcomers looking to ride the bioinformatics wave might be well served by a bioinformatics history lesson. If a successful company is a self sustaining, profitable purveyor of a valued product, then the truth is that to date there are few. By those parameters, the most successful bioinformatics business got its start more than 20 years ago in a lab in Madison, Wis., long before anybody used the term bioinformatics or “integrated R&D informatics platform.”
By staying small and focused, while others came and went, GCG, the Genetics Computer Group, has always been in the black. Still today, with as few as 50 employees, it brings in a significant chunk of its parent Pharmacopeia’s software revenue.
“GCG has become the de facto standard in the field,” says Bristol-Myers Squibb associate director of bioinformatics Dan Davison. “Still today the core bioinformatics software is the GCG suite.”
People such as Mike Stapleton, COO of Accelrys, a subsidiary of Pharmacopeia, which acquired GCG in August 2000, will argue that pharma companies are looking for providers of integrated platforms. But the fact remains: GCG is the only bioinformatics software company that has survived shifting market conditions and an unsuccessful acquisition. After 20 years, it is still in the black and its DNA analysis package is installed at more than 800 sites with more than 30,000 users.
John Devereux never intended to start a company. “I knew more about communism than capitalism when I started out,” the now-retired founder of GCG likes to say. Yet the DNA and protein sequence analysis programs he began writing more than 20 years ago at the University of Wisconsin at Madison became the most widely distributed commercial bioinformatics software in the industry’s history. And while a profitable bioinformatics company even today is a rarity, “we were never not profitable,” Devereux says. Not bad for a self-professed socialist. The secret to his success, he says, “is we never had any investors.”
Looking more Marxist than scientist in a black turtleneck and black wool trousers, Devereux, about to celebrate his 55th birthday, puts his career into a sociopolitical context. “Reagan made my career,” he says.
In 1977 Devereux dropped out of a PhD program in botany after five years to join what he considered to be the more exciting field of genetics. As a technician in the lab of molecular biologist Oliver Smithies, he wrote simple software programs to manipulate protein and DNA sequence. Reformat simply turned sequence into a format that was readable by computer. Reverse: put in one strand and get the other. But because nothing like it existed, he found himself sending copies to labs throughout the biology department.
Explaining how what was to be called UW-GCG (University of Wisconsin-Genetics Computer Group) was born, Devereux says, “When Reagan took office he cut a bunch of grants that NIH had already awarded.” Smithies’ lab was hit hard and Devereux was left supporting his computational work himself. With the university’s blessing, he began charging professors $2,000 a year for access to his software, thus creating the world’s first bioinformatics core facility.
It was 1982. Temple Smith and Michael Waterman had just published their eponymous algorithm. GenBank had completed its first year with 606 entries, and Blast wouldn’t appear for another eight years. “We weren’t trying to make money,” says Devereux. “In fact, we were forbidden to make money.” As a university-funded program, any remaining balance at the end of the year would be taken back.
GCG Goes Fungal
His success was not due to lack of competition. In 1980 at Stanford University another bioinformatics company was getting its start. Drawing on defense contract artificial intelligence research, Stanford computer scientists and molecular biologists created a suite of DNA and protein-data analysis programs and hosted them together with some databases on a computer. “We were using one of the very first Internet computers,” says Doug Brutlag, one of the Stanford biologists who went on to co-found IntelliGenetics. “Actually it was computer number 36 on the Internet.” Researchers worldwide would dial in to gain access to the tools and data.
“We found that we quickly had used up over half the CPU cycles of this computer that was supposed to be used for AI research,” says Brutlag. And many of the users were from pharma and biotech. Federal funding agencies were not too keen about the use of the public facility by commercial users. “So we decided to start our own company,” says Brutlag.
Over the next decade, as Apple and IBM introduced the PC, IntelliGenetics expanded its offerings by licensing in software. One package was a sequence-analysis tool called PCGene developed by a student at the University of Geneva as part of his PhD thesis. The student, Amos Bairoch, went on to create Swiss-Prot and co-found the proteomics company GeneProt. Another program called MacGene was developed by Case Western Reserve University grad student Manuel Glynias, now better known as CEO of NetGenics. But unlike GCG, IntelliGenetics quickly became a public company as a subsidiary of IntelliCorp, an AI company with defense contracts.
GCG’s software was so cheap that at $4,000 a pop, “people who decided to buy IntelliGenetics would buy GCG in the margins,” says Devereux. “And once we got it in there, it was better stuff, it was more efficient. We dislocated them in many places. We were like this insidious thing that would sneak in; no one knew we were even there,” Devereux laughs, joking that his software is like a bad virus or a fungus.“One of the things that made it commercially very stable was that once you get a substantial constituency in a company using any piece of software, you can’t get rid of it.”
All Is Fair…
IntelliGenetics, which was eventually sold by IntelliCorp to Amoco, gets the credit for making a private company out of GCG in 1989.
Over Thai food in a restaurant near the Madison campus, Devereux clearly recalls the conversation that turned him from academic to CEO of a profitable company. IntelliGenetics had been writing increasingly menacing letters claiming unfair competition and insisting that it would sue the university if it wouldn’t license out GCG’s software. When that wasn’t working, Intelli-Genetics turned to a different tactic, says Devereux. “They finally got their parent Amoco involved,” he says. An Amoco executive called the dean one evening, Devereux recounts, and “he said, ‘If you don’t license your software from this GCG facility to our subsidiary, IntelliGenetics, we’re going to withdraw all research funding at the University of Wisconsin. Do I make myself clear?’”
“Jesus, Bob, we’re being blackmailed,” Devereux remembers telling Dean Robert Bock. Naïve in the ways of business, Devereux sought the counsel of a lawyer with a letter from IntelliGenetics in hand. “After he read it he put it down on his table and said, ‘this is going to be the luckiest thing that will ever happen to you in your life. You just won the lottery.’”
The university wanted no stake in the new private company. So Devereux found himself as CEO and two-thirds owner of a company with more than $1 million in revenues from its built-in customer base. (Maggie Smith, Devereux’s business partner, owned the other third.)
The company grew as its chief product, known as the Wisconsin Package, expanded to contain more than 130 programs. And revenues continued to increase. There was constant cash flow and Devereux stayed away from investors. “It was just a beautiful, little, very slow-growing, always profitable company with very highly skilled people,” he says.
“GCG, even as a private company, was run like a software journal,” remembers BMS’s Davison. Devereux never called himself a CEO. He was more like a genial lab head, Davison says. And the company held seminars that attracted the giants of the field including Blast co-creators Stephen Altschul and Gene Myers, and NCBI developer Jim Ostell.
“It was a people-oriented affair,” says Devereux. “And to the extent that is possible in business, we tried to create a society.” Every year a pool of profits was split among employees. “There was absolutely no technical evaluation of people” in dividing the pot, he says. In the library of his stately new home in Madison, Devereux reaches for what at first glance looks like a wedding album. Each page features an eight-by-ten black-and-white glamour shot of a GCG employee. “This was my family,” he says.
But by the mid-’90s the competitive landscape began to change. “When we started, the field was wide open. It was like grabbing land in the Wild West,” Devereux recalls. Big guns were encroaching on the territory, the genome project was underway, and the bioinformatics gold rush had already begun in earnest.
Incyte, InforMax, MDL, MSI, Oxford Molecular, Pangea Systems, and PerkinElmer arrived on the scene. “We just didn’t have the resources to face that much firepower,” says Devereux. “Just a frivolous lawsuit and we’re out of business.”
Devereux and Maggie Smith were left with two choices: accept money from investors and ramp up, or sell to a bigger player. With only about $3 million in revenue they weren’t expecting to get much. So when Oxford Molecular CEO Tony Marchington offered $20 million cash in June of 1997, they took it.
Oxford Molecular, a public company based in Oxford, UK, quickly went south. Its strategy, it told investors, was to buy as many chemi- and bioinformatics companies as it could and offer a complete “solution” to pharma. The trouble was that there was no strategic integration among the disparate pieces. OxMol had also acquired Chemical Design Limited, MLR Automation, and Cambridge Discovery Chemistry — more than 25 software products in all. IntelliGenetics at this time was also an OxMol entity. It was trying to grow too much too quickly without any payoff strategy. “One thing that was very disappointing about getting to Oxford was recognizing how much all of it was based on press releases and only on press releases, because the investor was such an important part of the equation,” says Devereux.
Yet GCG chugged on. “We actually were able to grow GCG even with Oxford failing around us,” says Devereux. By 1998 GCG revenue increased by nearly 25 percent. Meanwhile OxMol as a whole was losing money quickly and had only $20 million in the bank. “We were flying high and they were siphoning it off. It made me mad,” Devereux says. So mad, in fact, that a year before OxMol crashed Devereux told the upper management at a high-level meeting: “Oxford is actually in a spooky stasis — like a cartoon character that runs off a cliff but doesn’t fall for awhile. We’ve already failed, but nobody’s told us.” Sure enough, OxMol blew its cash and in August 2000 sold its software business to Pharmacopeia.
Picking up the Pieces
Today the most active acquirer of bioinformatics assets is the Pharmacopeia subsidiary Accelrys. “We have pursued an acquisition strategy over the last few years and we’ll continue over the coming years,” says COO Stapleton. Pharmacopeia has been on a buying spree, picking up, among others, Molecular Simulations, Synopsys, and Oxford Molecular, and grouping them together under the umbrella of Accelrys. By bringing together bioinformatics, cheminformatics, molecular modeling, and consulting, the company hopes to become a one-stop shop for pharma and biotech R&D technology. If there is a bioinformatics company up for grabs, Accelrys will consider going for it. “Companies like Genomica, DoubleTwist — all of these we had a look at last year,” Stapleton says.
In 2000, the last full year reported, all of Pharmacopeia’s software products brought in $75.4 million in revenues. Only MSI contributed a bigger chunk to that number than GCG. “It’s still a must-have technology,” says Accelrys spokesman Jim Rivas.
Despite all the influx of investor cash into the industry, the Wisconsin Package is still one of the most profitable and widely used commercial bioinformatics software.
Whether Accelrys will be any more successful than Oxford Molecular remains to be seen. Despite Stapleton’s insistence that pharmaceutical companies prefer to get everything from one vendor, at least one pharma bioinformaticist disagrees. “There isn’t going to be a Microsoft in this business,” says BMS’s Davison. “There is no one solution. Every company does drug discovery differently,” he says. “I think of the bioinformatics industry as buying different components that meet your needs. … So the future is going to be companies that build small tools that play well with others.”
Other niche players that have followed the GCG approach — “focus on one thing and do it better than anyone else” — are finding success as well. One example is GeneCodes. The 22-person company in Ann Arbor, Mich., sells a single tool, Sequencher, that dominates the market for assembling BAC-sized and smaller DNA fragments. “We’ve had 34 consecutive profitable quarters,” says CEO Howard Cash, who insists he has no intention ever to sell his company or to broaden its offerings drastically. Meanwhile, in comparison, Lion lost Euro11.9 million in its latest quarter, has less than Euro16 million in the bank and has yet to see a profitable quarter.
“The cost of talent to have enough expertise in all areas of biology is so high that it’s hard to be profitable unless you’re building on one particular area of bioinformatics,” says Cash. “Only one company can hire Gene Myers to do frag assembly.”
Lion Bioscience’s bet, on the other hand, is that having a complete system is more important than any particular piece of software. “You have many companies … these days that come out with a piece of software and try to tell the world that they have the best software for a certain problem,” says CEO von Bohlen. “Even if that is true, the problem is not, ‘Is this homologous to that?’ or ‘Is this 3D structure better predicted than that 3D structure?’ The problem is, ‘How can I decide whether my R&D projects are likely to succeed or not?’ It’s a completely different question. It’s about, ‘Do we have patent rights on the chemicals? Are they metabolized into something toxic or not?’”
History has shown that the real losers are the companies that may very well make it as niche players, as GCG did, but instead bow to investor pressure and pitch a singularly focused product as a bigger solution. Industry insiders say that’s where Genomica went sour. “Genomica is one of those companies that had one software product and went out and tried to convince the world — and at least at the IPO it did — that it could build corporate-wide solutions,” says von Bohlen.
InforMax enjoyed great success when it just sold its desktop package VectorNTI. But sales of its enterprise system GenoMax have been disappointing.
For companies that oversell their offerings, the chance to be bought out is perhaps the best thing that could happen. “If I could do things over again,” says Pangea-turned-DoubleTwist founder Joel Bellenson, “we probably should have taken offers to buy us more seriously.”
Oxford Molecular’s downfall was pitching something it did not have; it sold disparate software products as a complete solution. “There was no strategic interaction among the pieces and that doesn’t survive very well,” says Michael Liebman, a pharma veteran and director of computational biology at the University of Pennsylvania Cancer Center.
Spinmeisters hurt the rest of the market as well. “They are pouring water into the wine,” says von Bohlen. “Everyone is talking about ‘strategic integration corporate-wide,’ using the same buzzwords.” The hyperbole can confuse the most seasoned veteran. “Even for me, it’s sometimes difficult to understand, OK, what is PR and what is real,” says von Bohlen.
Waiting for the Big Boom
While those gathered in the Pepper Hamilton penthouse are excited about investment opportunities in bioinformatics, the man who could be credited with christening the market sees it differently. “The business of bioinformatics has lost some of its tone. It’s become so driven by investors that it’s not a service anymore — it’s a speculation,” says Devereux. That speculation, yet to be proven, is that by spending billions on bioinformatics software platforms for early stages of the drug discovery pipeline, pharmaceutical companies will develop drugs more efficiently.
When Devereux started out, bioinformatics wasn’t about drugs at all. It was simply about making researchers’ lives a bit easier. Today, on the other hand, investors are betting that bioinformatics systems will directly be responsible for new drugs, and that the suppliers of the software platforms will make out like bandits. But with all the new tools they are buying, pharmaceutical companies are actually spending more money to get to each new drug target or lead than they were before.
Devereux points to a $125 million deal in 1993 as the event that changed the business of bioinformatics forever. “As soon as SmithKline bought the TIGR data through Human Genome Sciences, that transformed the whole industry and it’s never been the same,” says Devereux. Bioinformatics was no longer just a basic research tool. “It became driven by drug discovery,” he says. “From that point on the tail wagged the dog.”