There’s no right or wrong way to build an informatics system that makes the most of your genomic, proteomics, and chemical data, according to informatics professionals who convened in Scottsdale, Ariz., recently to share best practices in computational drug discovery.
The 40 or so attendees of Active Communications International’s Drug Discovery Informatics conference, January 23-25, were treated to an impressive speaker list and plenty of opportunities to interact and share ideas in the intimate setting.
Talks centered on the formidable obstacles pharmaceutical and biotech companies face in integrating their growing informatics capabilities into a unified — and effective — whole. Case studies from Aventis, Johnson & Johnson, Celera Genomics, Bristol-Myers Squibb, and other organizations highlighted the varying paths informatics leaders are taking to make the most of their current data while laying the groundwork for more to come.
Some firms, such as Wyeth-Ayerst Research and Bristol-Myers Squibb, are integrating their biological data first, with plans to add chemical data in the future. Peter Smith, director of discovery applications at Wyeth-Ayerst, discussed a new network-centric architecture his team built to handle biological data. Noting that “building one large informatics application to cope with the diversity, variability, and scale of biological data is undesirable,” Smith explained that the tool, called BioBench, uses a middle tier to break the tight coupling between client and server. The modular design of the system will allow Wyeth-Ayerst to add a chemistry component down the road, Smith said.
Another company to integrate from the biological domain towards chemistry is Bristol-Myers Squibb. Nathan Siemers, associate director of applied genomics at BMS, described a tool the company has been using for around six months to serve as a “clearinghouse” for gene-centric information. The repository, called GeneTracker, stores curated data for genes of interest to the company, along with target validation data and links to other relevant sources of information. Siemers said that BMS is currently working on linking chemical data into GeneTracker as well.
On the whole, pharmaceutical firms face more integration challenges than newer biotechs, such as Celera, Structural GenomiX, and AxCell Biosciences, who have had the relative luxury of building their informatics systems from the ground up. Pharmaceutical companies are faced with silos of different types of data generated by geographically dispersed research groups, along with the added integration complications that arise from a long M&A history and decades’ worth of data hiding in handwritten lab notebooks. Smaller firms who have launched in the past decade, however, can run in a leaner manner and have been able to keep their informatics needs in step with their growth.
As an example of this approach, Mark Adams, vice president of genome programs at Celera, provided a rare view of the company’s informatics strategy as it ramps up toward a full-scale discovery model. Adams said that the genome would remain at the core of Celera’s informatics infrastructure, allowing the company to integrate data from its growing list of proteomics, therapeutics, and chemistry activities into a single system.
Tod Klingler, senior director of bioinformatics at Structural GenomiX, described his company’s more focused architecture for structural biology, which merges experimental crystallographic data with a medley of computational data from ab initio predictions, homology modeling, virtual screening, and molecular dynamics techniques. Klingler said SGX has a chemistry lab in the works as well, and intends to integrate that data as seamlessly as possible into its system.
An interesting look at the informatics behind a database product was provided by AxCell Biosciences’ director of bioinformatics Roger Davis, who explained how AxCell’s bioinformatics algorithms, chemistry capabilities, molecular biology experiments, and screening activities are linked into a homemade laboratory information management system, which then serves as the foundation for the company’s ProChart database of protein-protein interactions.