Spatial Transcriptomics Analysis: Human Lymph Node (Visium HD)

Project Overview

This repository contains a complete analytical pipeline for Visium HD spatial transcriptomics data from a human FFPE lymph node sample. The workflow includes:

1. Quality Control
2. Normalization & Clustering
3. Cell Type Deconvolution
4. Cell-Cell Communication Analysis

Key biological insights include immune cell spatial organization, proliferative niches, and ligand-receptor interactions.

---

Repository Structure

├── data/  
│   ├── output_space_ranger_4-0-1/       # Raw Space Ranger outputs  
│   └── reference_dataset/                # scRNA-seq reference data  
├── results/  
│   ├── 01_qc_outputs/                    # QC plots and filtered data  
│   ├── 02_normalization_and_clustering/  # Clustering results  
│   ├── 03_cell_type_deconvolution/       # RCTD deconvolution outputs  
│   └── 04_cell_communication/            # CellChat interaction networks  
└── src/  
    ├── 01_spatial_QC.qmd                 # Quality control  
    ├── 02_normalization_and_clustering.qmd  
    ├── 03_cell_type_deconvolution.qmd    # RCTD deconvolution  
    └── 04_cell_communication.qmd         # CellChat analysis  

---

Key Workflow Steps

1. Quality Control (01_spatial_QC.qmd)

• Input: raw_feature_bc_matrix.h5 (10x Genomics Visium HD)
• Steps:
  • Filter low-quality spots (UMI < 30, genes < 40, MT% < 12.5)
  • Visualize QC metrics (UMI/gene distributions, spatial QC plots)
  • Cell cycle scoring and immune marker visualization (CD3E, BCL6, IGHG1)
• Output: Filtered Seurat object (spatial_qc_results.rds)

2. Normalization & Clustering (02_normalization_and_clustering.qmd)

• Methods:
  • BANKSY clustering with Leiden algorithm
  • Manual annotation using marker genes (e.g., CD3E for T cells)
• Output: Annotated clusters (Banksy_manually_annotated_object.rds)

3. Cell Type Deconvolution (03_cell_type_deconvolution.qmd)

• Reference Data: scRNA-seq from He et al. (2020) (lymph node + blood)
• Tools:
  • spacexr RCTD for spot deconvolution
  • Integration with BANKSY clusters
• Key Findings:
  • Myeloid/T cell spatial segregation
  • Rejected spots (~50%) enriched for non-immune cells (fibroblasts, ECs)

4. Cell-Cell Communication (04_cell_communication.qmd)

• Tool: CellChat with spatial constraints
• Analyses:
  • Global interaction networks (strength/count)
  • Pathway-specific visualization (e.g., MIF signaling)
• Output: Interaction probability matrices (cellchat_object.rds)

---

Dependencies

# Core packages
library(Seurat)       # v5.0.1
library(spacexr)      # RCTD deconvolution
library(CellChat)     # v1.6.1
library(BANKSY)       # Spatial clustering
library(scRNAseq)     # Reference datasets

# Supporting packages
library(tidyverse)    # Data wrangling
library(arrow)        # Efficient data storage

Usage

1. Run scripts sequentially:

Rscript src/01_spatial_QC.qmd  
Rscript src/02_normalization_and_clustering.qmd  
...

2. Outputs: Automatically saved to dated subfolders in results/.

Data availability

Data Availability Primary Data: 10x Genomics Visium HD (Human Lymph Node)

Reference scRNA-seq: scRNAseq::fetchDataset("he-organs-2020")