“Making your own microarrays is not the black art it used to be,” according to Dile Holton, a fact that PerkinElmer Life Sciences is counting on as it relaunches the ArrayInformatics software platform it inherited by acquiring Packard Biosciences last fall.
Holton, product manager for ArrayInformatics and QuantArray at PerkinElmer, said that the company is counting on sales of the software to labs that are just beginning to spot their own microarrays and are eager to buy the spotter, scanner, hybridization station, and software in one shot. “There’s been a change in the way people view microarrays,” said Holton: There are more choices, the technology to roll your own has improved, and prices have dropped, making customers “not as reluctant to buy the full package and spend $150,000-$400,000.”
In line with that assessment of the market, the ArrayInformatics platform was designed to meet the needs of researchers along each step of a gene expression microarray experiment — from probe selection, to array design, to image scanning, through to visualization and data mining of the results.
ArrayInformatics has had a long and circuitous journey to its current form. It was originally conceived within the life science division of GSI Lumonics, which was acquired by Packard BioScience in August 2000. Packard launched what Holton described as a “limited-feature” version of the platform just before PerkinElmer acquired the company in November 2001. “It really wasn’t the full product that we wanted. We just wanted to get ourselves positioned correctly” prior to the merger, said Holton, who has overseen development of the product since its conception at GSI. Now that the post-merger dust has settled, Holton said the “real release” of the platform is finally available through PerkinElmer.
Bells and Whistles
While earlier forms of ArrayInformatics focused on the image processing tied to Packard/ PerkinElmer’s scanners and other hardware, Holton said the new, “full-featured” release is built around a SQL Server relational database designed to store all the information associated with microarray experiments — a shrewd move for PerkinElmer, because the database architecture includes automated input points for the company’s SpotArray printers. Additional inputs to its liquid-handling and plate-handling hardware, scanners, and other instrumentation are planned for future releases.
But the platform was designed to be more than a centralized repository for raw data captured from PerkinElmer’s microarray hardware, Holton said. ArrayInformatics also provides a data mining function that guides users through normalization, filtering, clustering, and visualization via scatter plots and other options. Protocols are saved to track each step of the process so users can repeat a particular series of analytic steps on new or saved data. Users can also save lists of genes of particular interest that are linked to other sources of information about their function. Additional visualization and clustering features, such as principal components analysis, are planned for future versions of the product.
But it is these features of the platform’s capabilities that raise the biggest challenges for PerkinElmer. The market is already saturated with established players and startups alike offering gene expression analysis packages. Holton acknowledged that many users would likely prefer to use Spotfire, GeneSpring, or another software package for this end of the process, and said his team has “had discussions as to whether to open the architecture” to such tools, although this option is not yet available.
Michael Lombardi, who has been beta testing ArrayInformatics at the Center for Neurologic Diseases in Cambridge, Mass., for the past year, agreed that users are likely to prefer the data mining tools they’re accustomed to. However, he said, “the only problem with those is there’s no database, so you’re doing one slide at a time, essentially, and you’re on your own as far as cataloging after that.”
Warning that it takes “a bit of time to get started [on ArrayInformatics] in that a lot of the information needs to be entered by the users themselves,” Lombardi said that once the system is up and running it helps streamline the microarray experimentation process: “If you’re just getting started and you are going to be a high-throughput facility, this sort of thing would be very useful for keeping track of the incredible amount of information that’s generated for each and every slide.”
Indeed, the new microarray facilities are expected to be among the biggest customers for the software as they purchase it as part of a complete infrastructure. Perkin-Elmer is certainly relying to some degree on its large installed base of customers for its scanners and other tools as it markets ArrayInformatics, but Holton said the company has already sold the system to a number of labs along with spotters, scanners, and other hardware.
The “personal edition” of ArrayInformatics has a price tag of $20,000 and the “research edition,” designed for up to five users, runs around $38,000 for the software loaded on a dedicated high-end server. No list price for configurations of 10 users and above was available, but Holton said these installations would typically run between $60,000 and $100,000.
The next release of the software is expected in August. In addition to full integration with PerkinElmer’s instrumentation, the next version will be fully compliant with the MIAME standard, Holton said. He added that the company is looking to add a module for protein microarrays as it gears up to launch its own line of protein biochips later this year.