Nature Biotechnology 23, 1503 - 1508 (2005)
Published online: 6 December 2005; | doi:10.1038/nbt1164
Combined static and dynamic analysis for determining the quality of time-series expression profilesItamar Simon1, Zahava Siegfried1, Jason Ernst2
& Ziv Bar-Joseph2, 31
Dept. Molecular Biology, Hebrew University Medical School, Jerusalem, Israel 91120. 2
School of Computer Science, Carnegie Mellon University, 5000 Forbes Ave., Pittsburgh, Pennsylvania 15213, USA. 3
Department of Biology, Carnegie Mellon University, 5000 Forbes Ave., Pittsburgh, Pennsylvania 15213, USA.
Correspondence should be addressed to Ziv Bar-Joseph zivbj@cs.cmu.edu Expression profiling of time-series experiments is widely used to study biological systems. However, determining the quality of the resulting profiles remains a fundamental problem. Because of inadequate sampling rates, the effect of arrest-and-release methods and loss of synchronization, the measurements obtained from a series of time points may not accurately represent the underlying expression profiles. To solve this, we propose an approach that combines time-series and static (average) expression data analysis—for each gene, we determine whether its temporal expression profile can be reconciled with its static expression levels. We show that by combining synchronized and unsynchronized human cell cycle data, we can identify many cycling genes that are missed when using only time-series data. The algorithm also correctly distinguishes cycling genes from genes that specifically react to an environmental stimulus even if they share similar temporal expression profiles. Experimental validation of these results shows the utility of this analytical approach for determining the accuracy of gene expression patterns.
|
