Assessing gene-level translational control from ribosome profiling

Abstract

Motivation: The translational landscape of diverse cellular systems remains largely uncharacterized. A detailed understanding of the control of gene expression at the level of messenger RNA translation is vital to elucidating a systems-level view of complex molecular programs in the cell. Establishing the degree to which such post-transcriptional regulation can mediate specific phenotypes is similarly critical to elucidating the molecular pathogenesis of diseases such as cancer. Recently, methods for massively parallel sequencing of ribosome-bound fragments of messenger RNA have begun to uncover genome-wide translational control at codon resolution. Despite its promise for deeply characterizing mammalian proteomes, few analytical methods exist for the comprehensive analysis of this paired RNA and ribosome data.

Results: We describe the Babel framework, an analytical methodology for assessing the significance of changes in translational regulation within cells and between conditions. This approach facilitates the analysis of translation genome-wide while allowing statistically principled gene-level inference. Babel is based on an errors-in-variables regression model that uses the negative binomial distribution and draws inference using a parametric bootstrap approach. We demonstrate the operating characteristics of Babel on simulated data and use its gene-level inference to extend prior analyses significantly, discovering new translationally regulated modules under mammalian target of rapamycin (mTOR) pathway signaling control.

Fig. 1.

The Babel framework and analysis. (A) A schematic of the Babel framework in which ribosome profiling data are processed and aligned, and an NB-regression-NB model is developed in each sample to identify genes whose ribosome association is higher or lower (red and blue arrows, respectively) than expected (gray) from its mRNA expression level. Significant translationally regulated genes across all the samples of a given condition are determined (left) as are those that change significantly between conditions (right). (B) Although multiple candidate parametric and non-parametric regression forms were evaluated for estimating ribosome association based on mRNA abundance (plotted here as read counts in log-scale), the trimmed least squares approach was chosen (see text). (C) Errors-in-variables regression is justified by the intrinsic uncertainty of mRNA levels under the NB distribution where variability increases with increasing level of expression, demonstrated here for three genes with NB means of 100, 500 and 1000. The NB model of ribosome-given-mRNA counts is further necessitated by significantly greater variance than mean RPF counts across the distribution of mRNA abundance (inset)

Fig. 2.

Combining multiple tests. (A) Fisher’s method for combining two independent tests. (B) Our alternative approach based on the arithmetic mean of P-values. All equally significant pairs define a line passing through (d, 0) and (0, d), whereas more significant pairs correspond to a line with d’ < d (dark blue). Thus, the combined P-value corresponds to the area of the triangle. (C) One-sided P-values are used; therefore, the opposite triangle is used for pairs of consistently large P-values. (D) A practical example of the difference between approaches

Fig. 3.

Characterizing Babel performance. (A) The proportion of P-values below the given cutoffs as a function of the number of replicates for simulated data in which there are no translationally regulated genes. Results across simulated datasets indicate the type I error rates in Babel are as expected. (B) The power of the Babel framework for detecting translationally regulated genes (2–4-fold increase in ribosome association) as a function of the number of replicates at significance levels of 0.05 and 0.001 among genes in the middle 90% of the distribution of mRNA abundance

Fig. 4.

Analysis of mTOR inhibition. (A) The relationship between Babel-produced q-values for genes after mTOR inhibition with the adenosine triphosphate site inhibitor PP242 compared with the absolute value of translational efficiency as previously calculated (Hsieh et al., 2012). The rank of genes identified by Babel or by the original study is shown (inset), ranked by Babel significance. (B) Babel analysis of mTORC1 inhibition with PP242 identified several genes encoding distinct subunits of the eIF3 complex as selectively inhibited at the translational level, a signature that was absent upon rapamycin treatment or among the other eIF complexes. The eIF3 complex bound to the 40S ribosome is represented schematically [inset; inferred from references Masutani et al. (2007); Siridechadilok et al. (2005); Zhou et al. (2008)] with subunits colored as a function of the significance of their translational repression upon mTOR inhibition (as indicated)