15 OMICS A Journal of Integrative Biology Volume 10, Number 1, 2006 © Mary Ann Liebert, Inc. Meta-Analysis of Published Transcriptional and Translational Fold Changes Reveals a Preference for Low-Fold Inductions JONATHAN D. WREN and TYRRELL CONWAY ABSTRACT The goals of this study were to gain a better quantitative understanding of the dynamic range of transcriptional and translational response observed in biological systems and to examine the reporting of regulatory events for trends and biases. A straightforward pattern-match- ing routine extracted 3,408 independent observations regarding transcriptional fold-changes and 1,125 regarding translational fold-changes from over 15 million MEDLINE abstracts. Approximately 95% of reported changes were 2-fold. Further, the historical trend of re- porting individual fold-changes is declining in favor of high-throughput methods for tran- scription but not translation. Where it was possible to compare the average fold-changes in transcription and translation for the same gene/product (203 examples), approximately 53% were a 2-fold difference, suggesting a loose tendency for the two to be coupled in magni- tude. We found also that approximately three-fourths of reported regulatory events have been at the transcriptional level. The frequency distribution appears to be normally dis- tributed and peaks near 2-fold, suggesting that nature selects for a low-energy solution to regulatory responses. Because high-throughput technologies ordinarily sacrifice measure- ment quality for quantity, this also suggests that many regulatory events may not be reli- ably detectable by such technologies. Text mining of regulatory events and responses pro- vides additional information incorporable into microarray analysis, such as prior fold-change observations and flagging genes that are regulated post-transcription. All extracted regula- tion and response patterns can be downloaded at the following website: www.ou.edu/mi- croarray/oumcf/Meta_analysis.xls. INTRODUCTION E VALUATING THE SIGNIFICANCE of a biological response should start with what has been observed. The twofold rule of thumb for the significance of a transcriptional change is often used in high-throughput biology (Butler et al., 2003; Kudo et al., 2004; Li et al., 2004; Mix et al., 2004) and often questioned (Bas- sett et al., 1999; Albelda and Sheppard, 2000; Sabatti et al., 2002; Yang et al., 2002). It is obviously use- ful as a guideline for selecting a small set of transcriptional responders from within a large set of mea- Advanced Center for Genome Technology, Department of Botany and Microbiology, The University of Oklahoma, Norman, Oklahoma. 6103_02_p15-27 3/23/06 11:48 AM Page 15