A statistical test for conserved RNA structure shows lack of evidence for structure in lncRNAs

Abstract

Many functional RNAs have an evolutionarily conserved secondary structure. Conservation of RNA base pairing induces pairwise covariations in sequence alignments. We developed a computational method, R-scape (RNA Structural Covariation Above Phylogenetic Expectation), that quantitatively tests whether covariation analysis supports the presence of a conserved RNA secondary structure. R-scape analysis finds no statistically significant support for proposed secondary structures of the long noncoding RNAs HOTAIR, SRA, and Xist.

Figure 1. Independent substitutions on a tree can create confounding covariations

(a) Illustrative example showing a histogram of maximum covariance score per alignment, over 100 synthetic alignments simulated under three evolutionary models: no phylogeny/no structure constraint (red); phylogeny alone (grey); or phylogeny plus structure constraint (cyan). (b) Toy alignment (top left) with two independent substitutions (marked 1,2) on the same branch, resulting in an apparent pairwise covariation annotated by R2R (top right). R-scape simulated null alignments (bottom left) retain this confounding covariation signal, and it is judged insignificant (bottom right). (c) Toy alignment with five compensatory base pair substitutions (marked 1–5) showing a covariation pattern that is destroyed in the R-scape simulated null alignments, and thus judged significant.

Figure 2. Covariation analysis of known or proposed RNA secondary structures

(a) 5S rRNA as an example of an known structural RNA with significant covariation support. The plot (left) shows the expected null distribution (fit: black lines; data: black circles), compared to covariances observed for pairs in (blue) or not in (red) the annotated structure. (b) Covariation support for the Rfam annotated 5S rRNA structure, versus an alternative structure proposed by R-scape to include all significantly covarying pairs. Significant pairs (at E< 0.05) are highlighted in green. Coordinates are alignment column positions. Specific nucleotides are shown when their weighted frequency in the column exceeds 50%; black dots represent more variable positions. (c) On the positive y-axis, plot shows percentage of base pairs supported by covariation at three thresholds (red, E< 10^–5; orange, E < 0.05; grey, E < 10). Negative y-axis shows the number of additional significantly covarying pairs not in the proposed structure. Transfer RNA (tRNA), RNase P RNA, and the purine riboswitch (leftmost) are examples of functional structural RNAs with strong statistical support; three human DNA repeat elements (rightmost) are negative controls with no known RNA secondary structure constraint.