PHILADELPHIA--Leading players in the bioinformatics community will gather here early next month to begin tackling one of the field's thorniest challenges: developing common standards for an Interface Definition Language (IDL) that will allow industry and academic researchers to more easily share software objects and manipulate the growing mountain of genomic information and chemical, expression, and drug screening data.
Currently, such sharing and manipulation can be difficult because different labs, database curators, and software vendors have crafted customized informatics systems that are often incompatible. Many academic researchers have long been frustrated by the problem, but the emphasis on innovation in many grant-driven research labs and the newness of many technologies created few incentives for standardization.
In recent years, however, the rise of high-volume genomic laboratories, plus a new emphasis on bioinformatics in a pharmaceutical industry wary of sinking too much money into stand-alone software systems, has created powerful incentives to knock down barriers. Many bioinformatics professionals hope an emerging technology--the software architecture known as Common Object Request Broker Architecture (CORBA)--will allow software developers to do just that. First, however, software developers and users must agree to speak the same language when it comes to building software objects.
"The goal is to develop a common lingua franca that will allow the bioinformatics community to create sharable objects and build software from reusable components rather than create every line of code from scratch," explained David Benton of SmithKline Beecham Pharmaceuticals, who helped organize the meeting here. He has been joined in the effort by a diverse group of bioinformatics professionals, including Graham Cameron of the European Bioinformatics Institute, Tim Clark of Millennium Pharmaceuticals, Rainer Fuchs of Glaxo Wellcome, Juli Nash of Silicon Graphics, Eric Neumann of NetGenics, and Chris Overton of the University of Pennsylvania's Center for Bioinformatics.
In addition, a growing number of companies and nonprofit organizations have expressed support for the effort. They include Abbott Labs, Bristol-Myers Squibb, Chemical Abstracts Services, the Genetics Computer Group, the Genome Database at Johns Hopkins University Medical Institutes, Glaxo Wellcome, Incyte Pharmaceuticals, Millennium Pharmaceuticals, and Zeneca Pharmaceuticals.
The major goal of next month's meeting will be to begin a formal standards-setting process for the bioinformatics community within the Object Management Group (OMG), a nonprofit consortium based in Framingham, Mass., that was founded in 1989 to promote standards for object software. OMG has over 700 members, including both leading computer industry vendors and customers, who share an interest in modular object software and reusable code. The group has a web site at http://www.omg.org.
OMG is best known for CORBA, which many in the bioinformatics community believe will be the key to making sharable objects a reality. First released in 1991, CORBA defines an IDL that "allows applications to communicate with one another no matter where they are located or who has designed them," according to an OMG document. "It provides an infrastructure allowing objects to converse, independent of the specific platforms and techniques used to implement the objects. Compliance with the standard guarantees portability and interoperability of objects over a network of heterogeneous systems."
The introduction of CORBA has prompted a wide range of industries, from financial services to manufacturing, to pursue IDL standards through OMG's highly organized, task-force-driven process. Now the bioinformatics community is seeking to do the same. First, however, the group meeting here must develop a proposal to create a formal Special Interest Group (SIG) under one of OMG's three Technology Committees, which play a major role in determining standards. That proposal will probably be presented to an OMG committee meeting in Dublin, Ireland, September 22-25.
Eventually, researchers and industry officials told BioInform, the SIG could evolve into an OMG task force, a more permanent body that can issue requests for proposals to consortium members, evaluate responses, and decide on what OMG calls "specifications." While software developers are free to ignore the specifications, they do carry weight in the field, observers said.
"The reason we at Silicon Graphics are excited about this is that we've watched the OMG process bring together software providers, users, and academics involved in other industries, and then build a structure that provides leadership," Nash commented.
"We've been wanting to see this kind of effort for a long time," added NetGenics' Neumann, whose company has been using CORBA in developing its Synergy software framework for pharmaceutical and biotech companies. "We need to get consensus on at least a few critical pieces of interfaces. Then the next level of evolution can take place."
He acknowledged, however, that "in the past, it has been difficult to bring people together to create standards even when everyone agreed there would be a benefit. However, CORBA should alleviate this impasse by focusing on interfaces and not on implementation."
Others noted that some software developers are skeptical that a CORBA-centered system can currently deliver the high level of integration and interoperability that many are seeking; cure problems created by fundamental differences in database structures; or rein in the freewheeling software design culture.
"In theory, using CORBA should bring improvements," Overton observed. "In practice, there are going to be a lot of rough spots. We still have to demonstrate that you can build systems that can adequately handle large quantities of genomic data with CORBA."
Whatever its outcome, however, several bioinformatics professionals said the August meeting signals the maturation of the discipline. Once a highly theoretical pursuit confined to a few researchers' labs, the field is now becoming an applied science that industry is banking on heavily.
"This effort is a recognition that, to be efficient and productive in academia or industry, we can't all be reinventing the wheel," Overton noted. "Bioinformatics is evolving," added Millennium's Clark. "It is shifting from a craft discipline characterized by homegrown solutions designed for small-scale environments to an engineering discipline scaled up to handle high-throughput environments."
--David A. Malakoff