WASHINGTON, DC — GlaxoSmithKline will launch an internally developed microarray analysis package to the public for free, and in so doing will apparently become the first pharma to take such a step.
The decision is in stark contrast to strategies at other big pharmas, which have developed their own LIMS or analysis software for strictly internal R&D — a strategy that frustrates some academic researchers who desire that this kind of software be made open-source.
Kejun Liu, a senior statistician at GSK, said that the company would be releasing a final version of its PowerArray data-analysis and -visualization tool to the public in late September. A rudimentary version of the system is already available online, but Liu said that GSK needed to finish testing and documenting the system before the official first version is posted.
Liu spoke with BioArray News following his presentation at Cambridge Healthtech Institute's Total Microarray Data Analysis and Interpretation conference, held here last week.
"A big problem is that we have to get the documentation and testing project done as soon as possible," Liu said. "That's a joint effort from GSK, Sanofi-Aventis, and the National Institute of Statistical Standards. It's 95-percent done. After this project we will write a paper. The web site is already up and you already can download the trial version."
Liu described PowerArray as an "integrated analysis environment for high-dimensional data analysis and high-dimensional visualization." The system accepts data from a variety of sources, Liu said, including Affymetrix .cel files, raw data from proteomics platforms, and others.
Although the PowerArray system comes with features specific to the Affymetrix platform, like the ability to "take .cel files directly and view .cel images," Liu said that analysis and visualization could be done with data from any platform.
"Each -omic platform has some specific tools," he said. "After the data has been analyzed or processed by these specific tools, everything is the same in [our] analysis environment and the same high-dimensional visualization system."
Liu also said that PowerArray can enable researchers to work with more covariates from data, a robust principle component for generating heat maps, and the ability to validate samples.
During his presentation at the conference, Liu compared PowerArray to software developed by Spotfire. And although he insisted afterwards that GSK's product, which he co-developed as a postdoc at the National Institute of Statistical Science, wasn't comparable to the commercial informatics platform, he said that PowerArray could be set apart from commercial packages like Spotfire in a number of ways.
"Spotfire has a great visualization system, but has few analysis tools," Liu said. "Recently they just introduced their module for detecting differentiation, but you cannot have any covariates and only support up to two factors. PowerArray provides more flexibility on the analysis part."
Another strength Liu claims for PowerArray is its interface, which has been tailored for biologists rather than statisticians. He said he interviewed many biologists while he was creating the software at NISS in order to make it more biologist-friendly.
Commercialization a 'Complicated Issue'
Liu said that although PowerArray is a free tool, it could possibly compete against rival commercial platforms such as Spotfire and others. Yet despite this suggestion, Liu said that it is unlikely that GSK will derive any revenue from the product anytime soon: Distributing array data-analysis software for profit is not among his employer's priorities, and the software is co-owned by Liu, NISS, and GSK, which makes decision making more difficult.
"This is just for J&J. I cannot share software with you. I cannot share data with you. I can only share ideas with you."
"There are definitely enough markets for [selling it] and I think that PowerArray is designed well enough to be commercialized and be successful," Liu said.
"The ownership makes it a complicated issue," he went on. "In order to commercialize it I have to reach an agreement among NISS, GSK and myself. It's free because right now I cannot commercialize this, but I do want the academic users, and even industry users, to take advantage of the existing instrumentation and help them work on their products," he said.
Liu said that GSK and NISS have considered licensing PowerArray to other companies, but logistical problems were holding them back from reaping any kind of financial reward.
For Their Eyes Only
Even though GSK may not see it feasible to distribute the PowerArray software in the near future, the company has distanced itself from other pharmas in the space, which are content simply to show other academic and industrial researchers their lab information management systems and data-analysis and -visualization tools, but not sell it or release it gratis.
For instance, Xiang Yao, a statistician from Johnson & Johnson, told conference attendees last week that while his company is ready to share some of its secrets when it comes to array data management and analysis, it is not ready to release the software.
"This is just for J&J," he said after a conference participant asked if the LIMS he had described would become available to the public. "I cannot share software with you. I cannot share data with you. I can only share ideas with you," Yao answered.
Among the ideas Yao shared were how J&J manages to control microarray experiments being conducted and used at J&J labs in La Jolla, Calif., where most of its genomic resources are housed, and its drug-discovery labs in Belgium and Toledo, Spain.
He explained that because of the shortcomings of commercially available LIMS, J&J had to develop its own. "Affy's LIMS system is not a LIMS system. It's just data processing software," he said during his presentation.
Affy did not return a call for comment by press time.
Yao revealed during his talk that J&J primarily uses the Affymetrix and GE Healthcare's CodeLink platforms in its work. But instead of working with LIMS provided by Affy and GE Healthcare, J&J decided to develop its own LIMS.
"Our LIMS system is unique to our laboratory system flow," Yao told BioArray News after his presentation. Some of the features that caught the eyes of attendees included an administrative review of all projects before the experiments were given the go-ahead.
"[Since] most biologists don't really know how to design a microarray experiment very well, our [management] needs to know the project ahead of time," Yao said. He said that J&J had a few patents on its LIMS, but that he welcomed others to "borrow" from J&J.
Johnson & Johnson isn't the only pharma with a LIMS or data-analysis and -visualization system that won't be publicly available anytime soon. Bristol-Myers Squibb also has no plans to release its software, according to Petra Ross-Macdonald, a senior research investigator at BMS who was at the conference last week.
Ross-Macdonald echoed Yao's comments by saying that BMS wasn't in the business of providing software, and she recommended that academic labs should build their own rather than spend money to buy a LIMS or data-analysis system from Spotfire, Affy, or GE Healthcare.
"Every big lab has a lot of microarray data and they all choose to handle it differently," Ross-Macdonald said. She said that BMS had also used many open source platforms when constructing their bioinformatics system.
Still, in spite of the fact that conference attendees were getting a peek at J&J's bioinformatics programs, some were frustrated that there weren't plans from other pharmas to release their array data analysis and management tools.
"I don't know how them showing it is helping us. From the academic side we're looking for more open source stuff to come out," John Tobias from University of Pennsylvania's Penn Center for Bioinformatics, told BioArray News during the conference.
Tobias said that by releasing these highly developed tools, pharma could save money and headaches for a lot of people working in bioinformatics that can't afford to invest in $100,000 worth of software that's commercially available.
"If there's something out there that can cut costs and time for our customers, then we're interested in it," Tobias said.
— Justin Petrone ([email protected])