Customers are just beginning to adopt Oracle 10g — the most recent version of Oracle’s database — but the company is already tapping life science researchers for ideas on features to include in its next release, which bears the working name 11g.
As part of this effort, the company held its fourth Oracle Life Science Users’ Group meeting at its offices in Reston, Va., last week. Around 160 people registered for the two-day gathering to share best practices and tips, provide feedback on 10g along with wish list items for the next release, and peek under the hoods of other groups’ Oracle-based informatics infrastructures.
The meeting is a rare bragging opportunity for life science database builders, said Charlie Berger, senior director of product management for life sciences and data mining at Oracle. “We’re nearly ubiquitous in life sciences, but we’re almost invisible,” he said. At most life science informatics meetings, “people don’t say, “Let me show you what I built with Oracle.’”
The meeting has an additional purpose. The company has found that it’s not uncommon for customers to ask for new features that are already in the current version of the database, and this gathering offers an opportunity to evangelize a bit about the database’s numerous capabilities. Berger opened the meeting with an overview of options within 10g that were inclu-ded with the life science community in mind, such as the Blast algorithm that is now part of the data-mining suite [BioInform 09-15-03].
In addition, the company hopes to build awareness among its user base by expanding the Oracle Consulting Services team into the life science vertical market. “We’re still building the practice,” Berger said of the small consulting team, which draws from specialists across the company as well as a few domain experts. Oracle is currently offering its life science customers a free day of consulting services to help kick off the practice, Berger said.
Making a List
While 11g is still in its preliminary planning stages, the Oracle life science team already has a list of capabilities it would like to see in the final version, including ex-panded support for image data; additional gateways to third-party applications; support for additional data types and formats; expanded statistics capabilities; and support for more life science algorithms.
Oracle’s life science team is tiny in comparison to the thousands of developers who work on the product, but feedback from the life science user group does find its way into the pipeline. Blast is a case in point. “They may not even be aware of it,” Berger said, “but everybody out there using a banking system or anything else that runs on Oracle has the Blast capability built right into it.” Still, he stressed, there are no guarantees that life science wish list items make it into the database.
From most customer accounts, 10g is living up to Oracle’s claims. John Burke of UCB Pharma said that his company was one of the first to test-drive Oracle’s Blast capabilities and committed to 10g before it was even available. “I confess to a bit of nervousness at the time of the risks of vaporware,” he said. But so far, he said, he is pleased with what he’s seen. A slight miscommunication between Java and SQL tied things up for a bit, but that was solved with the help of the Oracle product managers and consulting team, he said. The verdict: Blast works in 10g.
Martin Widlake, another early adopter of 10g who heads up database services at the Sanger Institute, provided an honest assessment of the database. Sanger recently started moving from Tru64 to Linux, and while 9i ran great on a single Linux node, it was “no good” on a cluster using the company’s RAC (Real Application Cluster) technology, Widlake said. This capability, however, is “far better” in 10g, which was developed to have enhanced features for grid and cluster computing. One feature that has come in handy in the new version, he said, is the capability to include regular expressions with SQL. In one case, he said, plugging in a regular expression reduced the runtime for a particular function from one minute to two seconds.
Nevertheless, deployment of 10g is still very limited in the life science market. Most of the 10 vendors present at the meeting said they plan to support 10g, but few have done so to date. “We don’t run into customers who aren’t using Oracle,” said John Burns of MDL. “But most of our customers are not on 10g yet.” MDL plans to support 10g in the next release of its software, he said.
John Swallow, an engineering principal at Waters, said that the Oracle database is the “backbone” of the company’s informatics growth push. Waters recently acquired NuGenesis and Creon Lab Control for their software, and has renegotiated its license with Oracle so customers can run all of Waters’ informatics products under a single Oracle license agreement. All of Waters’ software products are currently undergoing a 10g upgrade, Swallow said.
Building a Loyal User Base
Oracle is still introducing new capabilities for 10g. Last week, the company posted a pre-release of Oracle Data Miner, a graphical user interface for its Oracle Data Mining tools, on the Oracle Technology Network website (http://otn.oracle.com/products/bi/odm/odminer.html). In addition, Rav Podowski, senior product manager for life sciences at Oracle, has developed a text-mining application for Medline that he plans to post on the OTN website “in a few weeks.” The application uses the Oracle Text feature to locate and classify Medline documents in an automated or user-defined fashion. Text mining, Berger said, is “a big frontier” for Oracle, and the company is committed to expanding its capabilities in that area.
Berger said that the company also wants to build up an “aftermarket” user community among its life science customers. A Sourceforge account has been set up for life science developers interested in sharing code they have developed for Oracle (http://sourceforge.net/projects/oraclelifesci/), and Berger said that he hopes to see customers bring their code to future user group meetings as well.