SOWhat is a high-performance R Shiny application designed to bridge the gap between automated single-cell analysis and manual cell type annotation. It provides an intuitive, interactive interface for exploring Seurat objects, identifying clusters, and refining annotations in real-time.
- Live UMAP Labels: Rename clusters in the annotation table and watch the UMAP plot update instantly.
- Feature Exploration: Search for individual genes and visualize their expression directly on the UMAP with the Feature Plot tab.
- Dynamic Cluster Seeding: New resolutions are automatically seeded with original cluster IDs to jumpstart your work.
- Resizable Interface: Balance your workspace with an interactive divider between plots and metadata.
- Multi-Tab Interface: Seamlessly switch between UMAP, Dot Plots, Module Scores, Clustifyr, and Markers.
- Signature Evaluation: Calculate aggregate module scores for gene sets using
AddModuleScore. - Species-Aware Presets: Integrated gene lists (Immune, Proliferation, etc.) that automatically detect and match the case convention (Human/Mouse) of your dataset.
- Reference-Based Identification: Load reference matrices from the
refmats/folder and compute spearman correlations. - One-Click Apply: Instantly transfer top-scoring cell types to your annotation table with the "Apply Top Cell Types" button.
- Smart Gene Matching: Intelligent case-insensitive gene name normalization ensures compatibility across different reference platforms.
- FindAllMarkers Integration: Run differential expression analysis directly from the UI.
- Interactive Heatmaps: Dynamically sized heatmaps that scale to fit your top markers without compression.
- Filterable Tables: Explore significant markers (P-adj < 0.05) in a responsive data table.
Ensure you have R (>= 4.4.2) installed on your system.
git clone https://github.com/MLKaufman/SOWhat
cd SOWhatThis project uses renv to ensure reproducible dependencies.
# Launch R in the project directory
R
# Inside R:
install.packages("renv") # If not already installed
renv::init() # To initialize the project
renv::restore() # To install all required packages from renv.lockLaunch the Shiny app directly from the root directory:
shiny::runApp()or from the command line:
Rscript -e "shiny::runApp()"- Upload Data: Use the sidebar to upload a Seurat object (
.rds). The app will initialize with a sample dataset if available. - Select Resolution: Choose the clustering resolution you wish to annotate.
- Annotate: Type your cell type identities into the Cluster Annotations table.
- Analyze:
- Use Dot Plots or Module Scores to verify markers.
- Use Clustifyr for algorithmic suggestions.
- Export: Click Save Annotations to download a CSV of your work, including reasons/comments for each annotation.
app.R: Core Shiny application logic.refmats/: Directory for.rdsreference matrices used by Clustifyr.docs/images/: Visual assets for documentation.renv.lock: Dependency lockfile.genelists/: Directory for gene lists used by the app.
Built with assistance from Google Antigravity and Gemini 3 Flash