Covariation Analysis - Usage Guide

Overview

Test whether a proposed or predicted RNA secondary structure is supported by evolutionary covariation using R-scape. R-scape scores compensatory substitutions across an alignment against a phylogeny-aware null and estimates statistical power, so the result is a three-way verdict (supports / rejects / cannot infer), not a simple pass/fail. This is the gold-standard validation that found no covariation support for the proposed HOTAIR, Xist, and SRA lncRNA structures.

Prerequisites

# R-scape (includes the CaCoFold structure predictor and R2R diagram output)
conda install -c bioconda rscape

# Input: a Stockholm alignment (deep and diverse), with a #=GC SS_cons line for -s.
# Build alignments with Infernal cmalign, RNAalifold, or from an Rfam SEED.

Quick Start

Tell your AI agent what you want to do: - "Is this conserved RNA structure actually supported by covariation?" - "Run R-scape on my alignment and tell me if the structure is real" - "Does my alignment even have the power to test structure?" - "Predict a covariation-supported structure from my alignment" - "Validate my SS_cons before I build a covariance model"

Example Prompts

Validate a Proposed Structure

"I have a Stockholm alignment with a consensus structure for a putative ncRNA. Run R-scape and tell me whether the covariation is significant or just phylogenetic."

"My lncRNA has a published secondary structure. Test whether it has evolutionary support."

Power and Interpretation

"R-scape found no significant pairs. Is that evidence against the structure, or does my alignment lack power?"

"How many diverse homologs do I need before a covariation test is meaningful?"

De Novo Structure

"I have no trusted structure. Build a covariation-supported consensus with CaCoFold."

Feed Downstream

"Validate this SS_cons with R-scape before I build a covariance model from it."

What the Agent Will Do

1. Confirm the alignment is deep and diverse enough for a covariation test to have power
2. Run R-scape with -s to test the given SS_cons, or --cacofold to build a structure de novo
3. Read the covarying-pair counts and per-pair power from the .cov and .power outputs
4. Return the three-way verdict (supports / rejects / cannot infer), not just a pair count
5. Pass a validated or CaCoFold structure to covariance-model building or thermodynamic folding

Tips

• Power makes the negative meaningful - "0 significant pairs" is only evidence against a structure if the alignment had power to detect covariation. A low-power negative is "cannot infer", not "rejects"; report which one.
• Depth and diversity - Covariation needs sequences that vary at paired columns while preserving the pair. Few or near-identical sequences give low power; gather more diverse homologs.
• Significance, not score - A positive covariation score is not validation; require the E-value to clear the target (default 0.05) against the phylogenetic null.
• Alignment quality - Misaligned columns destroy real covariation and can manufacture spurious signal; validate the alignment first.
• CaCoFold for de novo - When there is no trusted structure, --cacofold builds one from significant covariation (it can include pseudoknots), giving a strong consensus to seed a covariance model or compare against a thermodynamic fold.
• The lncRNA lesson - A thermodynamically plausible structure is not established until covariation is statistically demonstrated; many proposed lncRNA structures fail this test.
• Keep outputs tidy - Pass --outdir so the .cov/.power/R2R files do not land in the working directory.

Related Skills

• secondary-structure-prediction - Predict the structure whose conservation is then tested
• ncrna-search - Validate a custom CM's SS_cons here before building the covariance model
• structure-probing - Experimental evidence complementary to evolutionary covariation
• alignment/msa-statistics - Assess the alignment depth and diversity covariation needs
• phylogenetics/tree-io - The phylogeny underlying the covariation null