NEW YORK, March 9 – A Japanese tumorogenesis researcher arrived in New York on Wednesday to begin a mouse microarray project at Memorial Sloan-Kettering's Ludwig Center for Cancer Research only to find out that defects in the Affymetrix U74 mouse microarrays had stopped the project dead in its tracks.
“That whole plan has been six months in the making,” said Michael Marino, a researcher at the Ludwig Institute. “Today we find we can’t go ahead with the research – it’s just wasting time.”
Earlier this week, Affymetrix disclosed that its U74 mouse arrays contained non-functional probes. Y et, despite the problem, researchers say they remain committed to using Affy’s chips in their experiments.
“We’ve decided that Affymetrix is the best system out there in terms of pennies per gene analyzed,” Marino said.
Affymetrix has said it would offer replacements to the chips, in which 25 to 60 percent of sequence information is non-functional, within six weeks, and would offer researchers CD-ROMs in the meantime that specify which sequences are incorrect and enable re-analysis of the data that masks the incorrect sequence information.
“It will take us a month or two to be able to rerun those samples and data, with data analysis and storage issues,” said Andrew Brooks, director of the Functional Genomics Center at the University of Rochester Medical Center.
Brooks’ lab also makes its own oligo arrays, but he ruled out switching to homemade mouse arrays entirely. “We don’t have Affymetrix’s algorithm,” he said. “The thing with Affy is they have a patent on densities of genes, so they can screen large numbers of genes” at a time.
Researchers at Bristol-Myers Squibb’s Pharmaceutical Research Institute, one of the largest customers for Affy’s mouse arrays, remain similarly committed to Affymetrix's arrays in spite of this problem.
"In terms of confidence [in Affymetrix], what was good was we were able to validate that there was a problem with the chips,” said Paul Kayne, a senior research investigator at the institute's microarray facility. “If half the spots [had been] bad on every other chip, it would be difficult to sort out. But they were consistently wrong. This shows us that the overall technology is still working. They had bad input which led to a bad product."
Kayne said his lab would not have to re-do experiments entirely. Instead, they plan to take the correct sequence data points from the original experiments and re-analyze them. “The experiments are still usable,” he said. “There's just less there.”
Affymetrix has pointed out that 75 percent of the data on chips A and B of the three-chip set is usable, and 40 percent of the data on chip C is usable, for a total of 22,000 usable data points. The company has said some customers are saying they want to go ahead and use the chips with defective sequence because of the amount of usable data they can still obtain.
But not all researchers—especially those who already make their own microarrays—are convinced that Affymetrix is a reliable option.
Leming Shi, who does microarray analysis at the chemical company BASF, called the defective mouse arrays “a nightmare for customers.”
"Now there are quality control problems” in addition to the high price tag, he added.
Shi said his company makes cDNA mouse arrays in-house. And while this process involves more work for bench scientists at the beginning in fine-tuning the parameters, “we’re doing okay now,” he said.
Whether labs decide to stay with the Affymetrix arrays or not, researchers agree that a simple lesson can be learned from this debacle. "It's a reminder to all of us that we validate everything through some other means,” said Kayne.