# Rna2 Dfold

> Use when computing MFE structures, partition functions, and Boltzmann-sampled secondary structures within k,l distance neighborhoods relative to two reference structures for an RNA sequence.

- Skill: `vimalinx/rna2-dfold` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds add vimalinx/rna2-dfold`
- Raw SKILL.md: https://api.skillmd.com/api/skills/vimalinx/rna2-dfold/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: vimalinx (https://skillmd.com/u/vimalinx)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/vimalinx/rna2-dfold

---


# rna2-dfold

## Quick Start

- **Command:** `RNA2Dfold [OPTIONS] < sequence_with_two_structures`
- **Local executable:** `/home/vimalinx/miniforge3/envs/bio/bin/RNA2Dfold`
- **Full reference:** See [`references/help.md`](references/help.md) for complete options and model details

## When To Use This Tool

- Partition RNA secondary structure space by base-pair distance to two reference structures.
- Compare alternative folding neighborhoods around two candidate conformations for the same RNA.
- Compute MFE representatives and, with `-p`, ensemble statistics for each `(k,l)` distance class.
- Sample structures from specific neighborhoods with stochastic backtracking.

## Common Patterns

```bash
# 1) Compute MFE representatives and partition-function statistics
cat <<'EOF' | RNA2Dfold -p
GGGAAAUCC
(((...)))
((.....))
EOF
```

```bash
# 2) Restrict the explored distance range to both references
cat input.txt | RNA2Dfold -p -K 10 -L 10
```

```bash
# 3) Backtrack samples from one specific neighborhood
cat input.txt | RNA2Dfold -p --stochBT=100 --neighborhood=3:5
```

## Recommended Workflow

1. Prepare input: an RNA sequence plus two reference structures in dot-bracket notation
2. Run `RNA2Dfold -p` to compute partition function and Gibbs free energy for each k,l neighborhood
3. Use `--stochBT=INT` to generate Boltzmann samples from specified neighborhoods
4. Analyze MFE representatives and sampled structures across distance classes

## Guardrails

- Requires exactly two reference structures in dot-bracket notation as input alongside the sequence
- Distance bounds `-K` and `-L` must accommodate the desired neighborhood range
- `--stochBT` only makes sense together with `-p`