Why aren’t cancer microarray datasets archived more often?

A new study in PLoS ONE by Heather Piwowar, a postdoctoral associate affiliated with DataONE, Dryad, and NESCent, reveals interesting trends in the archiving of data underlying published microarray results.  From the press release:

By querying the full text of the scientific literature through websites like Google Scholar and PubMed Central, Piwowar identified eleven thousand studies that collected a particular type of data about cellular activity, called gene expression microarray data. Only 45% of recent gene expression studies were found to have deposited their data in the public databases developed for this purpose. The rate of data publication has increased only slightly from 2007 to 2009. Data is shared least often from studies on cancer and human subjects: cancer studies make their data available for wide reuse half as often as similar studies outside cancer.

“It was disheartening to discover that studies on cancer and human subjects were least likely to make their data available. These data are surely some of the most valuable for reuse, to confirm, refute, inform and advance bench-to-bedside translational research,” Piwowar said.

“We want as much scientific progress as we can get from our tax and charity dollars. This requires increased access to data resources. Data can be shared while maintaining patient privacy,” Piwowar added, noting that patient re-identification is rarely an issue for gene expression microarray studies.

Reference:  Piwowar, H. (2011). “Who shares? Who doesn’t? Factors associated with openly archiving raw research data.” PLoS ONE 6(7): e18657. doi:18610.11371/journal.pone.0018657

“In the spirit of the topic”, the data behind the study are publicly available in Dryad at doi:10.5061/dryad.mf1sd