RNA Structure Probing (bioSkills)

A Claude Code skill for the experimental half of RNA structure work: processing chemical-probing reads into normalized reactivity profiles, then using them to restrain — not replace — thermodynamic folding.

   
Type Claude Skill
Supplier GPTomics bioSkills (community OSS, MIT)
Availability GA — part of the bioSkills collection
Pricing Free / OSS (MIT) — ShapeMapper2 (Weeks Lab), ViennaRNA and SEISMIC-RNA are installed separately under their own licences
Capabilities Read/Write — Claude runs the skill’s workflow locally (Bash/Python), not as an MCP tool
Verified works · 2026-08-13
Security cleared · 2026-08-13 — GPTomics/bioSkills MIT, no external credentials

How to install

bioSkills is not an npm package — skills are plain markdown/code read directly by the agent. Clone the repo, then either run the installer for the whole category or copy the single skill directory.

  • Claude Code — clone and install via the bundled script: 
    git clone https://github.com/GPTomics/bioSkills
cd bioSkills
./install-claude.sh --categories "rna-structure"

    The installer copies matching skills into ~/.claude/skills/ (default target). Use ./install-claude.sh --list to preview the skills first.

  • Claude Code / other agents — copy just this one skill: 
    cp -r bioSkills/rna-structure/structure-probing ~/.claude/skills/

    (run from inside your clone — the previous step left you in bioSkills/; otherwise replace bioSkills/ with the absolute path of your clone, e.g. /Users/you/repos/bioSkills).

  • Prerequisites — the folding and plotting layer installs cleanly from conda/pip: 
    conda install -c conda-forge -c bioconda viennarna
pip install "matplotlib>=3.8" "pandas>=2.2" "numpy>=1.26"
  • ShapeMapper2 2.1.5+ — do not conda install shapemapper: the bioconda package of that name is ShapeMapper 1.2 (checked 2026-08-08), a different generation of the pipeline that will not satisfy this skill. Install ShapeMapper2 from the Weeks Lab release for your platform and put its directory on PATH. Unverified — the exact release filename and build steps were not confirmed from upstream this run; follow the install section of the Weeks-UNC/shapemapper2 README. ShapeMapper2 is Linux-only; on macOS run it under Docker or Singularity.
  • SEISMIC-RNA (optional, only for multi-conformation clustering) — see rouskinlab/seismic-rna for its own install path; the skill targets 0.20+.

What it does

Walks the probing experiment end to end, with the decisions that determine whether the resulting profile means anything:

  • Reagent choice — SHAPE reagents (1M7, NAI, 2A3) report on all four bases; DMS reads only A and C, so G and U must be masked to -999 rather than recorded as unreactive. Four-base DMS is treated as valid only when the protocol and analysis explicitly enable it.
  • Readout choice — mutational profiling (MaP) and RT-stop readouts need different scoring pipelines; the skill does not let one be processed as the other.
  • Sample design — three libraries, not one: MODIFIED (signal), UNTREATED (background), DENATURED (normalization reference).
  • Processing — ShapeMapper2 produces normalized per-nucleotide reactivity profiles with per-position depth and error.
  • Normalization — per-transcript 2–8% box-plot scaling.
  • Restrained folding — reactivities enter ViennaRNA as soft pseudo-energy restraints, via the Deigan model (m = 1.8, b = -0.6) or the Zarringhalam model.
  • Multiple conformations — per-read mutation data clustered with SEISMIC-RNA or DREEM when a single consensus structure is the wrong model.

QC thresholds the skill enforces: effective depth ≥ 5,000 (also ShapeMapper2’s --min-depth), untreated mutation rate < 0.5%, modified mutation rate roughly 1–10% (below that is undermodification, above it suggests degradation), and --max-bg 0.05. Nucleotides failing depth are reported as -999, never as zero.

Primary use cases: processing SHAPE-MaP/DMS-MaPseq libraries into reactivity profiles, folding a transcript with experimental restraints, detecting structural heterogeneity in a single transcript.

Notes

The framing rule is that reactivity probes constraint, not pairing — a protected nucleotide may be base-paired, or bound by protein, or stacked, or simply in a badly-sequenced region, and the skill insists the technical explanations be ruled out first. The corollary that most often goes wrong in practice: raw reactivities from different transcripts or experiments are not comparable, so they must never be pooled or compared before per-transcript normalization.

Upstream skill front-matter name is bio-rna-structure-structure-probing; upstream directory rna-structure/structure-probing. ShapeMapper2’s Linux-only constraint is the main practical obstacle for Mac users. Downstream and adjacent: ViennaRNA is the folding engine the restraints feed, Covariation Analysis provides the independent evolutionary line of evidence for the same structure, and ncRNA Search identifies which family a probed transcript belongs to.

Sources

Installed this tool?

Share feedback — install path, OS, errors, workarounds. The form opens with this tool pre-selected and a link back to this page.