usage-guide.md

Spatial Neighbor Graphs - Usage Guide

Overview

This skill covers building spatial neighbor graphs for spatial transcriptomics analysis using Squidpy. Neighbor graphs define which spots/cells are considered spatial neighbors for downstream analyses.

Prerequisites

pip install squidpy scanpy

Quick Start

Tell your AI agent what you want to do:

• "Build a spatial neighbor graph"
• "Create a KNN graph with 6 neighbors"

Example Prompts

Basic Graph

"Build a spatial neighbor graph for my Visium data"

"Create a Delaunay triangulation graph"

Specific Parameters

"Build a KNN graph with 10 neighbors"

"Connect spots within 100 pixels of each other"

What the Agent Will Do

1. Extract spatial coordinates
2. Build neighbor graph using specified method
3. Store connectivities and distances in adata.obsp
4. Report graph statistics

Tips

• Visium grid - Use coord_type='grid' and n_rings for hexagonal grids
• n_neighs - 6 is typical for Visium; adjust for single-cell resolution data
• Delaunay - Creates natural tessellation, good for variable density
• Radius - Use when you want fixed distance threshold