By Malorye Branca
August 13, 2002 | Recognizing the need for bioinformatics tools that work together, several software companies, including Accelrys Inc., InforMax Inc., LION Bioscience AG, TurboGenomics Inc., and SciTegic Inc., are starting to release new products designed to improve cross-product integration, thereby freeing up valuable time for bioinformaticians.
“We’ve been asking for years, ‘Where is it extendable, how can we incorporate what we’ve done?’” says Jochen Kumm, senior director of informatics at Davis, Calif.-based biotech firm Sagres Discovery.
But developers have put little effort into making their applications easy to integrate, partly because they prefer being paid to provide integration as a service. When the first generation of enterprise solutions was launched a couple of years ago, the products came with closed architectures, which probably dampened sales. “Our early GenoMax sold fairly well, but it didn’t sell well enough, and none of the comparable systems have sold well either,” says Steve Lincoln, chief scientific officer for InforMax, based in Bethesda, Md.
Users’ needs have changed as well. “Bioinformatics has evolved to where people are spanning multiple approaches,” says Lincoln. “They want to be able to wire together their own concoctions and deploy those out across the enterprise.”
TurboGenomics’ new TurboBench software is aimed at just that sweet spot. TurboBench lets researchers design their own workflows. These workflows run on TurboHub, an accelerator, which is also sold separately. “We want to make it as easy as possible for users to add new programs,” says Andrew Sherman, vice president of operations at the New Haven, Conn.-based bioinformatics company.
TurboBench comes with a set of prebuilt wrappers, available from a pull-down menu, which allow the easy integration of public domain software such as BLAST and PHRED. It has wizards (automated tools that build wrappers) that can create these for any other software. Once a wrapper is selected or built for a tool, that tool can be incorporated in future workflows and shared with other users.
SciTegic’s Pipeline Pilot also creates workflow. Users can integrate tools in this program using a Windows interface, four scripting languages (Perl, Python, VB script, and the company’s proprietary Pilot Script), and SOAP (simple object access protocol). “Our customers have not encountered any application that they can’t integrate,” claims J.R. Tozer, vice president of marketing for SciTegic, based in San Diego. The company has so far focused on chemistry informatics, but it is developing a version for bioinformatics to be launched by the end of this year.
Software suite builders are also taking heed of the desire for interoperability, making their own products more standardized and easier to integrate, and opening up their platforms to outside applications.
“We have to look at the reality of the situation,” says Steve Levine, senior director of strategic partnerships at Accelrys, based in San Diego. “Whenever [customers] choose our application, that is great. But whatever application they choose, we are there to help.” Accelrys is launching its new Discovery Studio (see New Products, page XX), and from the product’s inception, he says, “the idea has always been to deliver an open platform.” That includes standardizing Accelrys’ own software and making it easier to incorporate other companies’ tools into the system. Accelrys’ GCG (Genetics Computer Group) Wisconsin Package is one of the top tools other developers would like to integrate.
“There is a growing recognition by these vendors that sticking with a closed environment is not a feasible approach in the long run,” says Rainer Fuchs, vice president of research informatics at Cambridge, Mass.-based Biogen Inc.
“InforMax has been doing a fairly good job at addressing this,” says Sagres Discovery’s Kumm. “But overall, the industry is still weak in providing well-documented, easy-to-use interfaces.”
InforMax has enhanced its latest version of Vector NTI Suite (8.0) by giving it some of GenoMax’s best features, including workgroup capabilities as well as better, and easier, interoperability. “Our goal is to be able to quickly accommodate the vast majority of algorithms and methods for offering algorithms,” says Lincoln.
Commercial software is the toughest to integrate, because more information and formal agreements are usually required. “When we have a cooperative relationship with another vendor, it is much easier to create a simple interface,” says Michael Dickson, senior principal architect for LION Bioscience in Cleveland. Having already forged deals with several major database providers, including Celera Genomics, Incyte Genomics Inc., and Derwent Information Ltd., LION is now trying to hook up with “best in breed” application providers as well. “We have a strategic relationship with MDL [Information Systems Inc.] at present, but we would be very open to having a similar relationship with companies like Spotfire or InforMax, based on the needs of our customers,” says Dickson.
“People will forever be generating new ways of doing analysis,” says Sherman. “And they will always be asking, ‘How can I use that new program without throwing away everything that I already have?” As tools promising this capability head to market, it will be interesting to see whether they finally make clients happy.
Sidebar: Dollars for Data Integration
Users may seek interoperability in their software, but the big question for vendors is, “What are they willing to pay for?” The answer is “data integration,” according to a recent study conducted by Boston Consulting Group (BCG) for LION Bioscience.
The study suggests that pharmaceutical companies can save as much as $149 million on each new drug they develop by “integrating data and information across disciplines, pharmaceutical departments, processes, and drugs,” according to a press release from LION. The BCG analysis estimates that basic data integration could account for up to $87 million of this figure, and linking basic research data to clinical data could generate another $62 million.
The total potential savings attributable to effective informatics, on a per-new-drug basis, is $264 million. These figures are based on the assumption that the average cost of developing a drug is $880 million over 14.7 years.
There is unanimous agreement that researchers want better data integration. But most vendors find that clients’ requirements are extremely variable. As a result, data integration has become a bit of a boondoggle, always overpromised and underdelivered. “I have great respect for LION,” says Tim Clark, Millennium Pharmaceuticals Inc.’s vice president of informatics. “But there is only so much one company can do.”
LION’s CEO, Friedrich von Bohlen, points to the study as a vindication of LION’s decision to target seamless, value-generating data integration as its “raison d’etre.” But he also acknowledges the skepticism that is likely to emerge. “We’re very interested in how people react to the numbers,” he says. “I’m sure it will spur discussion.”
When reminded that he’s also generating expectations with the release of the study, von Bohlen is pragmatic. “The market drives strategy, and then there is only one place left to go,” he says. “You have to deliver.”