Attendees of O’Reilly & Associates’ first Bioinformatics Technology Conference anticipating sun-drenched desert weather were surprised by a freak snowfall that hit Tucson, Ariz., last week, but not to worry: As Tim O’Reilly assured everybody in his welcoming speech, “Bioinformatics is hot.”
Judging by the interest in the conference, it certainly is. More than 700 participants — ranging from bioinformatics pioneers to beginning practitioners to curious onlookers — convened in not-so-sunny Tucson to hear some of the world’s leading bioinformaticists speak about the current state of the field.
As the first in what will probably become an annual event, the O’Reilly conference organizers decided to take a survey-style approach, lining up speakers who could introduce newcomers to the essential elements of bioinformatics while provoking the interest of long-time practitioners. Co-chair Nathan Torkington, an O’Reilly editor, said the meeting was designed to address the same audience that purchased the publisher’s first two bioinformatics books, which debuted this year.
The books, “Developing Bioinformatics Computer Skills,” by Cynthia Gibas and Per Jambeck, and “Beginning Perl for Bioinformatics,” by James Tisdall, sold in record numbers and inspired the meeting.
As one might expect from an O’Reilly conference, there was a heavy emphasis on open source software development and related issues. Several panel discussions touched on accessibility to information in various forms, including biological data, intellectual property, and the scientific literature.
Tim O’Reilly told BioInform that he was impressed by the unified nature of the bioinformatics open source community “because they see themselves as part of the same project.” Unlike general-purpose software development, O’Reilly said, in bioinformatics, “the real project is not at the program level.”
Set up more like a computer trade show than a scientific meeting, four parallel tracks kept participants running between “Fundamentals,” “Analysis,” “Discovery,” and a fourth track that changed daily.
Plenary sessions presented a rare opportunity for speakers to discuss the art and science of bioinformatics in a more general sense than research-oriented meetings such as ISMB and Recomb.
In another departure from scientific custom, alternative rock standards accompanied keynote speakers on their way to the podium.
Day One: Bioinformatics 101
Ewan Birney of the European Bioinformatics Institute delivered the opening keynote, an overview of open source development efforts in the field. Birney stressed the complexity of the biological problems that underlie bioinformatics, warning curious computer scientists in the crowd that “if you don’t get excited about the data, I don’t think you’re going to stick to this field. Go back to e-commerce and design shopping carts.”
Lincoln Stein’s afternoon talk gave an overview of the field from a different perspective. Stein, of Cold Spring Harbor Laboratory, compared the current state of bioinformatics to Italy in the middle ages — a loose network of often-feuding city-states. Part science, part history lesson, the talk addressed the challenges of integrating bioinformatics data, software, and platforms in a unification process much like Italy experienced in the mid-19th century. Stein closed his talk by proposing a “code of conduct” for data providers to become “citizens of a nation.” (See sidebar below.)
Day Two: Putting Data To Work
The second day of the conference brought biological data into the picture. Terry Gaasterland of Rockefeller University began the day with a detailed discussion of the bioinformatics-based scientific method. She presented a step-by-step overview of a microarray experiment from the design stage through to analysis and discovery.
Filling in for Gene Myers, who was scheduled to speak in the afternoon but was unable to attend, Chris Hogue of the Samuel Lunenfield Research Institute first captured everyone’s attention with a South Park-esque animated short entitled, “Stephanie Finds Bioinformatics.” His talk covered the combined data set his group derived from two sets of protein-protein interaction data recently published in Nature and illustrated the value of bioinformatics in gaining biological knowledge from massive data sets.
Bioinformatics.org held its second annual meeting with an all-day track. The group, which hosts open source bioinformatics software and tools on its website, also presented its first Benjamin Franklin award for “promoting freedom and openness” to Michael Eisen of Lawrence Berkeley National Lab. Eisen was recognized for his development of the microarray clustering tools Cluster, TreeView, and ScanAlyze, as well as his efforts in support of the Public Library of Science initiative, which promotes free online access to the scientific literature.
Day Three: A Peek into the Future
The final day’s keynote talks addressed the future challenges and opportunities for bioinformatics. James Ostell of the NCBI delivered an overview of the history of the center with a look toward how it is evolving to cope with new types of information, such as scientific literature and images.
In the closing keynote, Lee Hood of the Institute for Systems Biology described how a multiple platform approach can be used to analyze biological systems. Ending his talk on a controversial note, Lee noted that “few academic institutions will be able to do real systems biology.”
A specialized I3C track gave attendees an update on the industry-wide consortium’s latest activities and provided a forum for I3C working groups to outline their efforts.
The conference ended with a round of five-minute “lightning talks,” followed by a moderated “town hall” session to gather feedback from attendees.
Lincoln Stein’s Bioinformatics Data Providers’ Code of Conduct
• A Web Page is an Interface: Include a page that explains how to link to your data.
• An Interface is a Contract: Since your links page will be “fossilized” in Perl scripts around the world, give fair warning if you make any changes.
• Choice is Good: Support as many ways to get to the data as possible.
• Allow Batch Download of your data.
• Use Existing Data Formats whenever you can.
• If you must create your own data format, Design Sensible Formats.
• Support Ad Hoc Queries: People will want to use your data in unexpected ways.