Researchers Clarify Outstanding Accuracy of GeneChip® Arrays’ Design and Manufacturing
10 Percent Data Error Attributed to Laboratory Experimental Design Flaws
Researchers studying statistical techniques to reduce data quality variations derived from microarray experiments have clarified that the 10 percent variation they found in data from 809 studies using more than 22,000 Affymetrix GeneChip® Human Genome U133A and Human Genome U133 Plus 2.0 arrays is due to poor quality data, not manufacturing quality.
Speaking on behalf of the scientists involved in the research, Dr. Rafael Irizarry said in this recently published paper it may be mistakenly implied that the cause of the bad data lies in the manufacturing of the arrays and that it should have stated more clearly that the source of error is due to “high throughput technologies riddled with bias and systematic errors; from rt-pcr to second generation sequencing.”
The focus of the research was to demonstrate some of the statistical techniques and methods the researchers developed to address variation associated with laboratory samples and procedures. Because the Affymetrix microarray technology is widely used, the Affymetrix platform presented the scientists with more published data with which to isolate experimental error and test their new normalization metric.
“The industrial nature of the Affymetrix GeneChip system and the large number of public samples makes the identification of bad data easier than for most other technologies," added Dr. Irizarry. "This paper demonstrates some of the statistical techniques we have developed for this platform. We expect to develop similar methods for other platforms but for now these useful tools are only available for Affymetrix."
Clarification of Findings
7 July 2011
Researchers Clarify Outstanding Accuracy of GeneChip® Arrays’ Design and Manufacturing
10 Percent Data Error Attributed to Laboratory Experimental Design Flaws
Researchers studying statistical techniques to reduce data quality variations derived from microarray experiments have clarified that the 10 percent variation they found in data from 809 studies using more than 22,000 Affymetrix GeneChip® Human Genome U133A and Human Genome U133 Plus 2.0 arrays is due to poor quality data, not manufacturing quality.
Speaking on behalf of the scientists involved in the research, Dr. Rafael Irizarry said in this recently published paper it may be mistakenly implied that the cause of the bad data lies in the manufacturing of the arrays and that it should have stated more clearly that the source of error is due to “high throughput technologies riddled with bias and systematic errors; from rt-pcr to second generation sequencing.”
The focus of the research was to demonstrate some of the statistical techniques and methods the researchers developed to address variation associated with laboratory samples and procedures. Because the Affymetrix microarray technology is widely used, the Affymetrix platform presented the scientists with more published data with which to isolate experimental error and test their new normalization metric.
“The industrial nature of the Affymetrix GeneChip system and the large number of public samples makes the identification of bad data easier than for most other technologies," added Dr. Irizarry. "This paper demonstrates some of the statistical techniques we have developed for this platform. We expect to develop similar methods for other platforms but for now these useful tools are only available for Affymetrix."
Competing interests
None declared