Bulkrna Splicing

---
# AUTO-GENERATED header from skill.yaml — do not edit by hand.
# Edit skill.yaml, then run: python scripts/generate_skill_md.py <skill_dir>
name: bulkrna-splicing
description: Load when summarising rMATS / SUPPA2 alternative-splicing output and identifying significant
  differential splicing events. Skip when you only have count-level DE (use bulkrna-de); splicing in single-cell;
  spatial data (currently unsupported).
version: 0.3.0
author: OmicsClaw
license: MIT
emoji: 🧬
tags:
- bulkrna
- splicing
- alternative-splicing
- PSI
- rMATS
- SUPPA2
requires:
- matplotlib
- numpy
- pandas
- scipy
---

# bulkrna-splicing

## When to use

Run AFTER rMATS or SUPPA2 has produced its splicing-event table — this
skill consumes that output (not raw alignments).  Computes per-event
ΔPSI (delta percent-spliced-in), flags events crossing significance and
ΔPSI thresholds, and groups results by event type (SE / A3SS / A5SS /
MXE / RI).

## Inputs & Outputs

<!-- AUTO-GENERATED from skill.yaml (interface) — do not edit by hand. Regenerate: python scripts/generate_skill_md.py <skill_dir> -->

**Inputs**

- File types: `.csv`

**Outputs**

- `tables/significant_events.csv`
- `tables/splicing_events.csv`
- `figures/dpsi_distribution.png`
- `figures/event_type_distribution.png`
- `figures/volcano_splicing.png`
- `report.md`
- `result.json`

## Flow

1. Load splicing event table.  Hard-fail at `bulkrna_splicing.py:365,368` on missing or invalid `--input`.
2. Validate the fixed input schema: must contain columns `event_type, gene, delta_psi, padj` (`bulkrna_splicing.py:153-155`).  No format detection — caller must pre-flatten rMATS / SUPPA2 output to this schema.
3. Filter by `--dpsi-cutoff` AND `--padj-cutoff`.
4. Group by event type; render distribution + volcano + bar plots.
5. Emit `tables/splicing_events.csv` (full) + `tables/significant_events.csv` (filtered) + report.

## Gotchas

- **This skill consumes the SPLICING TABLE, not BAM or FASTQ.**  Run rMATS or SUPPA2 upstream and feed their output here.  The wrapper does not perform splicing detection itself — feeding it BAM files raises a parser error or silently produces an empty result.
- **`--dpsi-cutoff` is the ABSOLUTE value of ΔPSI.**  Default `0.1` keeps events with `|ΔPSI| ≥ 0.1`, including both inclusion-up and inclusion-down.  Set to `0` to keep all directionally significant events.
- **Input schema is fixed: `event_type, gene, delta_psi, padj` (with optional `pvalue` and `event_id`).**  The script does NOT auto-detect rMATS vs SUPPA2 column conventions — if your input uses rMATS's `IncLevelDifference`/`FDR` or SUPPA2's `dPSI`/`pval` natively, rename columns first or the loader will silently drop your data.
- **Event-type breakdown depends on the upstream tool's classification.**  rMATS reports SE / A3SS / A5SS / MXE / RI as separate files; SUPPA2 uses an EVENT field.  Concatenate / re-label these into a single `event_type` column before feeding the skill, or the breakdown bar chart under-counts.

## Key CLI

```bash
python omicsclaw.py run bulkrna-splicing --demo
python omicsclaw.py run bulkrna-splicing \
  --input rmats_se.csv --output results/
python omicsclaw.py run bulkrna-splicing \
  --input suppa2_events.csv --output results/ \
  --dpsi-cutoff 0.2 --padj-cutoff 0.01
```

## See also

- `references/parameters.md` — every CLI flag and tuning hint
- `references/methodology.md` — rMATS vs SUPPA2 format conventions, event-type taxonomy
- `references/output_contract.md` — exact output directory layout
- Adjacent skills: `bulkrna-de` (parallel: gene-level DE, complements exon-level splicing), `bulkrna-enrichment` (downstream: pathway view of splicing-affected genes via gene-symbol mapping)