Three studies with potentially profound implications for genetics and disease research are published in the May issue of Nature Methods. They reveal important new insights into how investigators can greatly enhance the quality and reliability of data obtained in experiments conducted with microarrays.
Every once in a while, a new technology dramatically changes the way research is conducted and establishes an entirely new experimental paradigm. The microarray has proven to be just such a breakthrough, enabling detailed examination of gene expression and regulation at a previously inconceivable scale.
Microarray experiments use tiny chips containing thousands of precisely arranged gene fragment probes to measure the expression levels of individual genes in a biological sample. Each experiment consists of many steps, however, and controversy is raging over the accuracy and repeatability of these experiments. This is a serious issue for a scientific community that is working hard to build centralized depots of consolidated genetic data. With microarrays seeing ever-increasing use in 'fishing expeditions' for genes linked to various diseases and to cancer, scientists are understandably concerned by the prospect that these studies may be fundamentally flawed.
The three articles -- representing the collective effort of seventeen different labs -- attempt to shed new light on this debate by identifying obstacles to reproducibility in microarray experiments. All three come to essentially the same conclusion: most microarray platforms can be made to perform reliably, but the key to consistency is overcoming the so-called ?lab effect? by standardizing best practices for sample preparation, experiment design, and data collection across the research community.
Gavin Sherlock, in an associated 'News and Views' feature, summarizes the findings. "The three papers in this issue provide a cautionary tale for microarray research, but also a reason for optimism... they demonstrate that it is possible to perform microarray experiments that are reproducible between labs and across platforms, provided standard methodologies are adopted for best performance."
John Quackenbush (Dana-Farber Cancer Institute, Boston, MA, USA)
Rafael A. Irizarry (Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA)
Brenda K. Weis (Toxicogenomics Research Consortium, National Institute of Environmental Health Sciences, Research Triangle Park, NC, USA)
Gavin Sherlock (Stanford University Medical School, Stanford, CA, USA)
(C) Nature Methods press release.
Message posted by: Trevor M. D'Souza