To err is human. Scientists are keenly aware of this, and the multistage nature of microarray experiments provides ample opportunity for the human aspect of experimental error. Robotics could offer a solution, as well as potential for greater experimental efficiency.

febit's GENIOM automates many steps of microarray experiments.

“It used to be that the only people who thought about automation were the drug screeners, the diagnostic folks, people who had to do things hundreds of thousands of times,” says John Blume, vice-president of assay and application product development at Affymetrix of Santa Clara, California. “But a lot of early-stage discovery-type work now happens at the level of hundreds of things at a time.”

Affymetrix uses a modular approach, developing units that automate individual experimental stages. Several other companies have taken a similar path after finding that many customers had already taken matters into their own hands. “A lot of them have already adopted automation infrastructure,” says Kevin Meldrum, director of genomics marketing at Agilent Technologies, Santa Clara, California. “They don't want to have to go out and buy a completely new system.”

San Diego-based array manufacturer Illumina offers an 'arrays of arrays' format inherently designed for higher-throughput, and so further automation is a lower priority, although the company offers an AutoLoader that can process the scanning of up to 32 BeadChips — containing up to 256 arrays in total — in 24 hours.

Chip-builder and service provider NimbleGen Systems, based in Madison, Wisconsin, has taken an opposing approach. “Our mission is to make everything automated,” says Emile Nuwaysir, vice-president of business development. “So that analysis is monitored and quality controlled by humans, but not run by humans.” He projects near-complete automation of the full process by the end of the year.

febit Biotech in Heidelberg, Germany, brings this philosophy to the benchtop with the GENIOM, a system that fully automates most processes, including array synthesis, hybridization and analysis. Current GENIOM arrays are limited to 6,000 features but will soon expand to 15,000, and Peer Stähler, vice-president of marketing and sales at febit, believes that the instrument's speed and flexibility offer benefits for rapid experimental probe design and optimization. “You can easily import and synthesize the results of other people's bioinformatics,” he says, “and you can export any capture probe that you have evaluated.”

M.E.