Analyzing Spatial Transcriptomic Landscapes of Human Breast Cancer

This project analyzes spatially resolved gene expression data generated by 10x Genomics’ original Visium Spatial Gene Expression platform. By integrating this data with a single-cell RNA-seq reference, I could map spatial cell type compositions within human breast cancer tissue.

---

📘 Project Overview

Starting from a slice of human breast cancer tissue, spatial transcriptomics was performed using the 10x Genomics Visium platform. The goal was to uncover spatial gene expression patterns and identify the regional distribution and co-localization of cell types across the tissue section.

A matched single-cell RNA-seq reference from a breast cancer atlas (Wu et al., 2021) was used to deconvolute spatial gene expression and estimate cell type proportions for each spatial spot.

---

📊 Analysis Tasks

Task 1: Preprocessing and Spatial Clustering

• Load filtered UMI and gene expression matrices into a Seurat object
• Perform quality control to remove low-quality spots
• Normalize using SCTransform
• Dimensionality reduction via PCA
• Clustering and annotation of spatial spots
• Visualization via UMAP and Spatial FeaturePlots
• Identification of spatially variable genes

Task 2: Spatial Deconvolution with SPOTlight

• Prepare a reference single-cell RNA-seq dataset:
  • Downsample to ~100 cells per major cell type (9 types)
  • Identify highly variable genes
• Use SPOTlight (v0.99.8) to deconvolute spatial data and assign estimated cell type proportions to each spot
• Visualize spatial cell type proportions and regional distributions

---

💡 Notes

• Raw data is not publicly available due to client ownership and confidentiality.
• Some example outputs plots are organized by task in the output/ folder.
• This project is designed for both reproducibility and clarity.

---

📬 Contact

Author: Nasim Rahmatpour Email: nasimrahmatpour1@gmail.com GitHub: (https://github.com/nasimbio)