R-scape

RNA Structural Covariation Above Phylogenetic Expectation

R-scape looks for evidence of a conserved RNA structure by measuring pairwise covariations observed in an input multiple sequence alignment. It analyzes all possible pairs, including those in your proposed structure (if you provide one). R-scape uses a null hypothesis that takes phylogenetic correlations and base composition biases into account, which can be sources of apparent pairwise covariation that are not due to conserved RNA structure.

The input is an RNA multiple sequence alignment in Stockholm format, optionally (usually) with an annotated consensus secondary structure. The structure may include pseudoknots. Only the first alignment in the file is analyzed; if you submit a Stockholm file containing more than one alignment, the others are ignored.

The output is a list of covarying pairs that are statistically significant at a given E-value. The E-value (or threshold for the number of false positives expected). For each covarying pair, the output also reports the inferred number of substitutions that occurred at these two positions in the phylogenetic tree, and the estimated statistical power for R-scape to detect a significantly covarying base pair when that much variation is present.

R-scape has two different modes of operation which determine how it calculates E-values. One mode analyses all possible pairs equally. The other mode requires a given consensus structure and performs two independent covariation tests: one on the set of proposed base pairs, the other on all other possible pairs. For either mode, R-scape can calculate a structure based on the observed covariations using the CaCoFold algorithm.

To see an example, you can download the Rfam seed alignment for the purine riboswitch (RF00167) - or any other Rfam RNA alignment - and upload that file to R-scape.

Analyze an RNA alignment

Find base pairs with statistically significant covariation support.

Optionally, propose a structure compatible with all significantly covarying base pairs.

Choose a mode

Evaluate region for conserved structure

All possible pairs are analyzed equally in one single test. If a consensus structure is provided, that structure is ignored in the covariation test, but it is visualized with the significant covarying pairs highlighted in green.

preferred use:

This option is most appropriate if you're trying to determine if a conserved structure exists.

Predict new structure

All possible pairs are analyzed equally in one single test. A structure is predicted and visualized with the significant covarying pairs highlighted in green.

preferred use:

This option is most appropriate for obtaining a new consensus structure prediction based on covariation analysis.

Evaluate given structure

Requires that your Stockholm file has a proposed consensus structure annotation. Two independent covariation tests are performed, one on the set of proposed base pairs, the other on all other possible pairs. The given structure is visualized with the significant covarying pairs highlighted in green.

preferred use:

This option is most appropriate for evaluating how well an independently proposed consensus structure is supported by covariation analysis.

Improve given structure

Requires that your Stockholm file has a proposed consensus structure annotation. Two independent covariation tests are performed, one on the set of proposed base pairs, the other on all other possible pairs. A new consensus structure is predicted and visualized with the significant covarying pairs highlighted in green.

preferred use:

This option is most appropriate for using covariation analysis to improve your current consensus structure.

Download

Current source code distribution: rscape.tar.gz

Documentation

R-scape manual: R-scape_userguide.pdf

Publications

A statistical test for conserved RNA structure shows lack of evidence for structure in lncRNAs. E Rivas, J Clements, and SR Eddy. Nature Methods 14:45-48, 2017.

Estimating the power of sequence covariation for detecting conserved RNA structure. E Rivas, J Clements, and SR Eddy. Bioinformatics, in press, 2020.

RNA structure prediction using positive and negative evolutionary information. E. Rivas. bioRxiv preprint, 2020.