Skip to content

usage-guide.md

Latest commit

 

History

History
54 lines (36 loc) · 1.3 KB

usage-guide.md

File metadata and controls

54 lines (36 loc) · 1.3 KB

Spatial Deconvolution - Usage Guide

Overview

This skill covers estimating cell type composition in spatial transcriptomics spots using reference-based deconvolution methods like cell2location, Tangram, and RCTD.

Prerequisites

# cell2location
pip install cell2location

# Tangram
pip install tangram-sc

# RCTD (R package)
# install.packages('spacexr')

Quick Start

Tell your AI agent what you want to do:

  • "Deconvolve my Visium data using this scRNA-seq reference"
  • "Estimate cell type proportions in each spot"

Example Prompts

Basic Deconvolution

"Run cell2location on my spatial data"

"Use Tangram to map cell types to spatial spots"

Visualization

"Plot cell type proportions spatially"

"Show the dominant cell type in each spot"

Comparison

"Compare deconvolution results with marker gene expression"

What the Agent Will Do

  1. Load spatial and reference scRNA-seq data
  2. Find shared genes and preprocess
  3. Train deconvolution model
  4. Estimate cell type proportions per spot
  5. Visualize results

Tips

  • Reference quality - Better reference = better deconvolution
  • Marker genes - Help Tangram; cell2location learns automatically
  • N_cells_per_location - Adjust based on tissue/platform
  • Validation - Check correlation with known marker genes