Sashimi Plots - Usage Guide

Overview

Visualize splicing events as sashimi-style plots showing per-sample read coverage and splice junction arcs labeled by read counts. Tools differ in input handling, group aggregation, and customization: ggsashimi (publication overlays), rmats2sashimiplot (rMATS-aware), MAJIQ-VOILA (LSV posteriors interactive HTML), leafviz (leafcutter Shiny), Jutils (tool-agnostic), pyGenomeTracks (multi-track figures).

Prerequisites

# Python / CLI
pip install pandas
conda install -c bioconda ggsashimi rmats2sashimiplot pygenometracks

# R for leafviz (leafcutter is on GitHub, not Bioconductor)
# devtools::install_github('davidaknowles/leafcutter/leafcutter')

# MAJIQ-VOILA bundled with MAJIQ V3 (majiq.biociphers.org)

# Jutils
conda install -c bioconda jutils

Quick Start

Tell your AI agent what you want to do: - "Create sashimi plots for the top 20 differential splicing events from rMATS" - "Visualize a specific exon-skipping event with samples grouped by condition" - "Generate publication-quality sashimi with intron shrinking and per-condition aggregation" - "Make MAJIQ VOILA interactive HTML for browsing LSV posteriors" - "Plot splicing alongside ChIP-seq tracks for the same locus"

Example Prompts

Single Event Visualization

"Plot a sashimi for chr17:43094000-43125000 (BRCA1 region) with control vs treatment groups, intron shrinking, and matched y-axis scales."

Batch Plotting from Differential Output

"Iterate ggsashimi over the top 25 significant rMATS events; output per-event PDFs with flanking 500nt context."

MAJIQ VOILA

"Run voila on MAJIQ deltapsi output to generate interactive HTML browser of LSV posteriors and splice graphs."

leafviz

"Generate leafcutter Shiny app for interactive cluster-level browsing with NMD annotation."

Multi-Track Figures

"Use pyGenomeTracks to combine RNA-seq coverage tracks with H3K4me3 ChIP-seq for the same locus."

Tool-Agnostic Heatmaps

"Use Jutils to create a unified heatmap of significant events across rMATS, leafcutter, and MAJIQ output."

What the Agent Will Do

Choose visualization tool based on the upstream differential analysis (rMATS, leafcutter, MAJIQ) or general region request
Configure sample grouping with consistent colors per condition
Apply intron shrinking, fixed y-scales, and replicate aggregation flags
Iterate over significant events for batch plotting
Generate publication-ready PDFs/SVGs with annotation tracks

Tips

Use --shrink for genes with large introns (TTN, brain genes) to keep exons visible
--fix-y-scale is essential for cross-group comparisons; otherwise auto-rescaling exaggerates differences
Aggregate replicates with -O 3 -A mean_j to reduce clutter while preserving variance via alpha
Limit to 3-4 groups per figure; more becomes hard to read
Include 200-500 nt flanking exons for splicing context
For MXE events, plot both alternative exons; otherwise only half of the event is visible
VOILA requires MAJIQ build output (splicegraph.zarr in V3; the V2 splicegraph.sql is deprecated) plus the .voila file
IGV sashimi is for interactive ad-hoc inspection, not figures

differential-splicing - Identify events to plot
splicing-quantification - Context for PSI values
data-visualization/genome-tracks - Multi-track figure design
data-visualization/ggplot2-fundamentals - ggsashimi customization