In a bid to attract more "statistically savvy" users to its visualization and analytics platform, Spotfire is adding a new software package for microarray analysis to its suite of DecisionSite tools.
The product, called DecisionSite for Microarray Analysis, or DSMA, is meant to complement DecisionSite for Functional Genomics, or DSFG, which has been Spotfire's primary product for the microarray-analysis sector since it was first released in 2001 [BioInform 07-23-01].
Brendan Gibson, DSMA product manager, said that the new product, which begins shipping this month, is targeted toward users who are "more statistically oriented" and "more interested in low-level or upstream processing of microarray data" than the current customer base for DSFG.
Targeted toward biologists, DSFG handles files that have already been processed and normalized using Affymetrix's MAS 5.0 software or other statistical approaches. DSMA now allows users to conduct those upstream processing steps within the DecisionSite framework. The software can process Affy CEL and CHP files or files from two-channel technologies, Gibson said.
Gibson said that while DSFG "is and remains focused on functional interpretation" with a user base comprising biologists "doing more functional interpretation of the data," the intended user for DSMA is "interested in trying different statistics or new algorithms to process the CEL files, and interpret what's going on at the experimental level."
"You have to spend more time looking at the data, but at least it's in the same package, so you can save time because you can compare the two different analyses and decide which one is better for your particular situation."
Spotfire developed DSMA in collaboration with statistical software firm Insightful as part of a partnership the firms announced in June. The new package includes the RMA (robust multichip analysis) and GCRMA (GeneChip RMA) normalization algorithms for processing CEL files, as well as multifactor ANOVA (analysis of variance), principal components analysis, and several clustering methods. DSMA can also be extended to include custom statistical methods written in other packages, such as SAS, R, or the R-based Bioconductor libraries.
Interest in alternatives to MAS 5.0 has been growing in the microarray-analysis community, and Gibson said the company developed DSMA to appeal to "power users" looking for a wider range of statistical analysis options but unwilling — or unable — to tackle S-Plus, SAS, or R on their own.
"A lot of people are aware of Bioconductor," Gibson said, but the open source project "is not very easy to use." He added that a number of Spotfire customers have been "very vocal" about adding RMA and GCRMA, in particular. "They know the algorithms are out there, but they're not sure how to get started," he said. DSMA was designed to provide similar functionality as high-end statistical packages, but with a point-and-click interface that Spotfire users are comfortable with.
In addition, the software is integrated with DSFG, so customers can upgrade their current license to obtain the new features. DSMA is also available as a standalone product. The company did not provide pricing information.
Gibson acknowledged that DSMA users are "definitely a smaller subset" of Spotfire's DSFG customer base. "We're seeing about 10-15 percent of the DSFG user community who are interested in these features," he said. Nevertheless, he noted, this vocal minority tends to be made up of "thought leaders" within the market who influence longer-term usage patterns.
Aura Pérez, an assistant professor at Case Western Reserve University who has been beta-testing DSMA, said that the software has proved helpful so far, "because it gives you one more way of looking at the data."
Pérez noted that this can have its drawbacks as well as its benefits, however. "Of course, then you have to spend more time looking at the data, but at least it's in the same package, so you can save time because you can compare the two different analyses and decide which one is better for your particular situation." Before trying DSMA, she said that she used either MAS 5.0 or GeneTraffic for the upstream analysis, and used DecisionSite "just to look at the endpoints after I got the values."
As for whether DSMA produces better results than MAS 5.0 or other approaches, Pérez said it's too soon to tell. "The results are definitely going to be different, just because you're using different analyses. The question that you have to decide is which one is better, and nobody really knows which one that is right now."
— Bernadette Toner ([email protected])