Most pharmaceutical informatics departments are unwilling to entrust any of their software development duties to an outside party, making the high-end tools developed by expert bioinformaticists unavailable to their bench biologists. But Sergey Ilyin, bioinformatics group leader at Johnson & Johnson Pharmaceutical Research & Development, said that a little flexibility allowed J&J’s bioinformaticists and biostatisticians to successfully collaborate with Spotfire, and to develop code that researchers across the organization were actually able to use.
Ilyin said that the J&J R&D group, based in Spring House, Pa., has just wrapped up the first phase of the collaboration and that it has has “exceeded our expectations.” The initial project focused on microarray data, but Ilyin said that work has already expanded into real-time PCR and will likely move into proteomics, and possibly pathway analysis.
The partnership began when it became apparent that the scripts developed by the company’s biostatisticians were too unwieldy for J&J bench scientists to use when analyzing their microarray data, Ilyin said. In-house tools developed with the R and S-Plus statistical languages, for example “require a lot of training [to use], and generally not everybody in an organization has the skills necessary to perform that type of analysis, even though it gives outstanding results and allows them to extract the most value out of the data,” he said.
Working with Spotfire, the J&J team created “guided applications” where they could plug these statistical scripts into the “scaffold” of the familiar Spotfire user interface. “This way, we can take scripts developed by our best biostatisticians and then create an application so that the [end-user] no longer needs to work with the script, and data can be analyzed efficiently by using basic commands,” Ilyin said.
The resulting microarray analysis applications removed one of the bottlenecks to broader use of the technology within the company, Ilyin explained, because “the previous limitation was that only selected people were trained to properly use these tools, and as a result, utilization of the technology was not complete.”
The arrangement also overcame a common criticism of both custom and out-of the-box bioinformatics solutions — a lack of flexibility once the software is installed. The plug-and-play nature of the Spotfire-based system allows Ilyin’s team to add or remove scripts whenever necessary, leaving the choice of analysis method up to the user. Ilyin said that the solution also offered cost savings, since now they can rely on free code, such as R, and a single commercial software platform. “The only expense comes in terms of paying Spotfire,” he said, at a cost “which is relatively low compared to other packages.”
Ilyin admitted that he was initially “somewhat reluctant” to hand over any portion of the bioinformatics team’s software development to an outside party, but was pleased with the results of Spotfire’s work, which “exactly matched the expectations of the internal scientists.” Implementation of the guided applications took around two months, he said.
The J&J team has already begun to work with Spotfire developers on the real-time PCR application, Ilyin said, and is looking toward proteomics as the next area of focus. “At the rate this process is going,” he said, “I expect that we will try to use the solution for pathway analysis as well.”