Meta-analysis and meta-regression of transcriptomic responses to water stress in Arabidopsis

Abstract

The large amounts of transcriptome data available for Arabidopsis thaliana make a compelling case for the need to generalize results across studies and extract the most robust and meaningful information possible from them. The results of various studies seeking to identify water stress-responsive genes only partially overlap. The aim of this work was to combine transcriptomic studies in a systematic way that identifies commonalities in response, taking into account variation among studies due to batch effects as well as sampling variation, while also identifying the effect of study-specific variables, such as the method of applying water stress, and the part of the plant the mRNA was extracted from. We used meta-analysis, the quantitative synthesis of independent research results, to summarize expression responses to water stress across studies, and meta-regression to model the contribution of covariates that may affect gene expression. We found that some genes with small but consistent differential responses become evident only when results are synthesized across experiments, and are missed in individual studies. We also identified genes with expression responses that are attributable to use of different plant parts and alternative methods for inducing water stress. Our results indicate that meta-analysis and meta-regression provide a powerful approach for identifying a robust gene set that is less sensitive to idiosyncratic results and for quantifying study characteristics that result in contrasting gene expression responses across studies. Combining meta-analysis with individual analyses may contribute to a richer understanding of the biology of water stress responses, and may prove valuable in other gene expression studies.

Keywords: Arabidopsis thaliana; drought response; gene expression microarray; meta-analysis; research synthesis; transcriptomics; water stress.

Figure 1

Meta‐analysis of the weighted mean of log expression ratios (i.e. weighted mean effect size, $\widehat{\mathrm{\mu }}$) of gene expression responses across water stress experiments, sorted by distance of the 95% confidence interval from zero (a), or distance of the weighted mean expression ratio from zero (b). Genes with confidence intervals that overlap zero in (a) are also sorted by the distance of the mean from zero. Each gene is indicated by a dot ($\widehat{\mathrm{\mu }}$) and a horizontal line (95% CI). Genes with significantly (FDR < 0.05) positive values of $\widehat{\mathrm{\mu }}$ (up‐regulated) are colored red; genes with significantly negative values of $\widehat{\mathrm{\mu }}$ (down‐regulated) are colored blue, and genes with values of $\widehat{\mathrm{\mu }}$ that are not significantly different from zero are colored gray. For genes with 95% CIs that overlap zero in (a), and for all genes in (b), the dots for $\widehat{\mathrm{\mu }}$ tend to merge into lines.

Figure 2

Differences in gene expression responses between shoots and roots identified using meta‐regression. Weighted means of log expression ratios ($\widehat{\mathrm{\mu }}$) for each gene are summarized in roots (red) and shoots (green), which are horizontally adjacent to each other. The value of $\widehat{\mathrm{\mu }}$ for each gene is indicated by a dot, with a horizontal line encompassing the 95% confidence interval (CI). The top half of the plot includes 167 genes that are up‐regulated in shoots and down‐regulated in roots, while the bottom half of the plot shows 76 genes that are down‐regulated in shoots and up‐regulated in roots.

Figure 3

Comparison of genes identified via t tests in individual studies and genes identified in the meta‐analysis. (a) Distribution of the number of studies (contrasts) that each differentially expressed gene was identified in, according to individual t tests performed in each study. (b) Venn diagram showing the intersection between genes identified by a t test in at least one study, and those identified by the meta‐analysis.

Figure 4

Examples of gene expression responses to drought that were detected by the meta‐analysis and meta‐regression, but not by the individual t test contrasts. (a) Expression response of JAM3 across ten individual experiments, with t test P values for each experiment shown above each bar. $\widehat{\mathrm{\mu }}$ and its confidence interval from the meta‐analysis (FDR < 0.001) are indicated by dashed and dotted lines, respectively. (b) Expression responses of EER5 across nine individual experiments. Plot as in (a), except that the results for experiments on shoots and roots are colored green and brown, respectively, and μ_shoots and ${\mathrm{\mu }}_{\mathrm{roots}}$ and their confidence intervals from the meta‐regression (Q_M = 15.3, FDR = 0.002) are indicated by separate dashed/dotted lines for shoot and root experiments.