Bio-IT World attendees this year had a number of new bioclusters to choose from, as several vendors were demonstrating hardware products bundled with pre-installed suites of bioinformatics software.
Microway introduced its first pre-configured bioinformatics product at the show with its Navion-HTX Personal Bio-Cluster, which uses AMD Opteron 64-bit processors and can be configured in groupings of four 1U nodes. The cluster runs Blast, HMMer, ClustalW, and NAMD.
A four-node system costs around $15,000, according to Bill Applegate, a Microway sales rep. Applegate said that people are "looking for smaller clusters" lately, and are moving away from centralized computational resources toward departmental, and even personal, HPC systems.
This philosophy certainly holds sway at Orion Multisystems, which was showing off its new DS 96 96-node personal deskside system, which looks like a bulked-up CPU and plugs into a standard wall outlet. Orion officially launched the system several weeks ago [BioInform 05-02-05], but was promoting its "personal supercomputing" contest, in which the research team who submits the most "innovative" computational problem can win up to two weeks, or 3.68 CPU years, of computing time on a DS-96. Winners will be selected in August.
Fujitsu was demonstrating its own version of a biocluster, with a twist. The group has just begun testing the first versions of its BioServer to be shipped to the US. The system, which has been under development since the company launched its US-based BioSciences unit last year [BioInform 05-03-04], will eventually include 1,920 processors into a single six-foot rack with 2.5 Tflops of computing power. Since it uses Fujitsu's FR-V low-power processors, the chips can be packed more densely and the system runs cooler than other clusters, according to Michael McManus, vice president of the BioSciences group.
The company was demonstrating the capabilities of the system for running so-called "high-throughput docking" as a complement to high-throughput screening. McManus described the system as "an appliance" for virtual screening. It runs three docking algorithms with various degrees of granularity: ShapeDock, the fastest one, which can narrow a library of millions of compounds down to several thousand; FastDock, a more precise docking program that can narrow the search down to the hundreds; and then a version of the Gromacs molecular dynamics algorithm for calculating free energy of binding.
Although Fujitsu hasn't disclosed pricing for the BioServer yet, McManus said that a single 2U case with 128 processors would come in "under $100,000."
Another bundled bioinformatics solution at the show came in the form of Unleashed Informatics' DogBox, a Sunfire v20z Dual Opteron server running the Blueprint Initiative's SeqHound data warehouse of more than a dozen bioinformatics databases. The system is self-updating, and includes information from GenBank, RefSeq, SwissProt, PDB, GO, Pfam, Entrez Gene, and other sources. DogBox includes a Sun StorEdge FC3511 array with 3 TB of storage.
The starting price for the system is $28,500.
IBM to Deploy Clinical Genomics Platform for Utah Population Study
IBM, the University of Utah Health Sciences Center, and LineaGen announced a collaboration to create an informatics infrastructure that will combine genetic and phenotypic information from several large patient databases in the state.
The system, which will be built upon IBM's Clinical Genomics technology and services offering, will integrate a total of around seven million patient records from the Utah Population Database and the Utah Genetic Reference Project Database located at the University of Utah Health Sciences Center. LineaGen, a non-profit company formed to manage commercial access to the databases, will oversee the project for the university.
In a statement, Mike Leavitt, HHS secretary and former governor of Utah, said the collaboration "will accelerate medical discovery through genetic research and help make personalized therapies a reality."
Michael Paul, chief operating officer of LineaGen, told BioInform that the Utah Population Database includes more than seven million "person records," which each include birth certificates, driver's license information, family histories, diagnostic records, pharmacy records, imaging data, and pathology results. In all, he estimated the database includes a total of more than 20 million individual records.
In addition, he said, another million medical records will come from the University's Health Sciences Center. Genetic information will come from the Utah Genetic Reference Project Database, which was created for the Centre d'Etude du Polymorphisme Humain, or CEPH, genetic mapping project, and then later used for the HapMap project.
Mike Svinte, vice president of information-based medicine at IBM, said that the system IBM plans to build for the Utah project will be similar in some ways to its previous clinical genomics projects at the Mayo Clinic and the Cleveland Clinic. One difference, he said, is the "breadth and depth" of the data for this project. Another is the Utah project's intention to focus initially on autoimmune disorders.
Paul stressed that an important part of the project will involve building "layers of access" to ensure patient privacy.
Future phases of the collaboration will involve developing analytical tools to mine the data and study complex genetic disorders. Svinte said that the collaboration will draw from several projects underway at IBM's Watson Research Lab for this stage of the project, and the tools will likely touch on data mining and pattern recognition, as well as emerging areas like molecular modeling and image analysis.
"The value will come after the initial data is integrated," he said.
BII Ramps up Commercialization Efforts
Ian Humphrey-Smith, CEO of the Biosystems Informatics Institute, told BioInform that the research initiative is quickly ramping up its commercialization efforts.
Last week, BII announced that it had created a commercialization arm, called Turbinia, and had secured exclusive rights to distribute constraint-based modeling software developed at Moscow State University and several other Russian research institutes.
Humphrey-Smith said that the most promising area for this software is in microbial research, and that commercialization efforts will likely focus on the bioprocess and biofermentation markets.
Meanwhile, Turbinia is also co-developing proteomics software with BII partner Nonlinear Dynamics, has another partnership underway with Lucidyx in the area of data integration, and yet another project with an undisclosed Eastern European research group on biomarker analysis.
BII was created last year with £4.5 million in funding from the UK's Department of Trade and Industry. The goal of the initiative has always been commercialization, Humphrey-Smith said. "We're not just a bunch of academics playing." The DTI funding provides an approximate "two-year window" for reaching profitability, he said.
BII currently employs 12 people, and is advertising for 12 more positions, including chief scientific officer, chief operating officer, head of systems biology, head of protein-protein interactions, and several research-level positions.
Gene-IT Launches GenomeQuest 2.0
Gene-IT launched a new version of its GenomeQuest sequence-analysis system, which keeps track of user queries and then issues an e-mail alert when new information is available on a given sequence.
Gene-IT has targeted GenomeQuest at intellectual property professionals over the last several years, but GenomeQuest 2.0 should also appeal to research scientists looking to reduce the amount of time they spend searching multiple resources for new sequence information, according to CEO Ron Ranauro.
The company has secured a "couple dozen" customers for GenomeQuest over the last year and a half, Ranauro said. The updated system is expected to appeal to users who require up-to-the-minute information on their sequences of interest, using information in the public genome repositories, as well as the international patent databases, which are quickly becoming large sequence data resources in their own right.
"In the last year, there was more sequence data available in patents than there were in the last 22 years," Ranauro said.
There is no limit to the number of queries that the system can store, and users can "dial up or down" the level of sequence homology that they require for their search.
GenomeQuest 2.0 is shipping now, Ranauro said.
Vivisimo Expands Life Science Search Platform Beyond PubMed
Vivisimo, which launched its ClusterMed technology for clustering PubMed search results a year ago [BioInform 04-05-04], launched a new version of its system that allows users to cluster search results for multiple internal and external data sources.
The new platform, called Velocity for Life Sciences version 4.2, uses the company's Clustering Engine technology to search once against multiple databases, and bring back relevant information that is automatically categorized into hierarchical folders.
Donald Taylor, senior director of life sciences solutions at Vivisimo, said that the system allows users to "digest a hundred times more information in less time."
The new system includes a "meta-alert" feature through which users receive alerts from several sources in one combined e-mail. Velocity also uses NCBI's Entrez LinkOut feature to provide full-text access to licensed journals and databases.
Taylor said that an undisclosed pharmaceutical company is currently using the system for more than 2,000 researchers.
A free live demo of Velocity using 20 publicly available data sources is available at http://biometacluster.com/.
Pricing for the system is based on the number of users and resources, Taylor said.
InforSense, Spotfire Work with GSK on Application-Building Project
InforSense and Spotfire announced that they are working together with GlaxoSmithKline to create a framework for GSK chemists to write research applications on the fly.
The system will use web services that GSK has been writing for a number of its internal systems, along with Spotfire's DecisionSite visual analytics platform, and InforSense's KDE workflow environment.
David Hadfield, senior vice president and general manager of Spotfire, said that the collaboration is a sign that pharmaceutical companies are moving away from "hard-wired" enterprise-wide informatics infrastructures that focus on "vertical" analytical processes, such as gene expression analysis, in favor of flexible frameworks that enable researchers to write and deploy applications as they are needed, and then reuse them as necessary.
Yike Guo, CEO of InforSense, said that researchers at GSK will use InforSense's KDE interface to build customized research workflows using several hundred GSK applications. These applications can then be deployed through the Spotfire DecisionSite visual analytics system, which is already in use by thousands of scientists at the company.
The system is currently used only for cheminformatics, but Guo said that it would be applicable to any other research area within the enterprise.
TurboWorx, HP Team on Cluster Platform
TurboWorx said that it will integrate its SmartGrid workflow and cluster-management software with HP compute clusters.
SmartGrid includes TurboWorx Builder, TurboWorx Enterprise, and TurboWorx ClusterManager.
HP's cluster platforms include both HP ProLiant IA32 Xeon DP and AMD Opteron hardware configurations.