Single-Cell RNA-Seq Analysis Project: Peripheral Blood Mononuclear Cells (PBMCs)

Project Overview

This repository documents an introductory single-cell RNA sequencing (scRNA-seq) data analysis pipeline for Peripheral Blood Mononuclear Cells (PBMCs). The project aims to demonstrate fundamental steps in scRNA-seq analysis using Python and the scanpy library.

Dataset

The analysis is performed on a publicly available dataset:

• Title: "5k Peripheral Blood Mononuclear Cells (PBMCs) from a Healthy Donor (v3 chemistry)"
• URL: https://www.10xgenomics.com/datasets/5-k-peripheral-blood-mononuclear-cells-pbm-cs-from-a-healthy-donor-v-3-chemistry-3-1-standard-3-0-2

Software and Dependencies

This project is developed using Python and relies heavily on the scanpy library for single-cell data analysis. The environment is managed using Conda.

Key libraries include:

• scanpy
• numpy
• pandas
• matplotlib
• leidenalg
• python-igraph
• scikit-learn (dependency for some scanpy functions)

To set up the environment, use the provided environment.yml file:

conda env create -f environment.yml
conda activate scrnaseq_env

Analysis Steps

The 01_initial_data_exploration.ipynb Jupyter notebook covers the following core scRNA-seq analysis steps:

1. Data Loading and Initial Quality Control (QC):
   • Loading raw count data.
   • Filtering out low-quality cells (based on mitochondrial content, total counts, and detected genes).
   • Filtering out unexpressed genes.
2. Normalization and Log-Transformation:
   • Normalizing gene expression counts to account for sequencing depth differences.
   • Log-transforming the normalised data.
3. Feature Selection:
   • Identifying and selecting highly variable genes (HVGs) for downstream analysis.
4. Data Scaling:
   • Scaling gene expression values to ensure genes contribute equally to distance calculations.
5. Dimensionality Reduction:
   • Principal Component Analysis (PCA): Linear reduction to identify major sources of variation. [Number of PCs chosen: 15]
   • UMAP (Uniform Manifold Approximation and Projection): Non-linear reduction for 2D visualization of cell similarity.
6. Clustering:
   • Leiden Algorithm: Graph-based clustering to group cells into distinct populations based on their similarity. [Resolution used: 0.5]. [Number of clusters identified: 13]
7. Cell Type Annotation:
   • Identification of cluster-specific marker genes using differential gene expression analysis.
   • Manual assignment of biological cell type labels to each cluster based on canonical markers.

Key Findings (Summary)

• Initial quality control effectively removed low-quality cells.
• Dimensionality reduction revealed distinct cell populations in the UMAP embedding.
• Leiden clustering identified 13 distinct cell clusters.
• Manual annotation, based on canonical marker genes, assigned preliminary identities to these clusters, including major immune cell types such as CD4+ T cells, CD8+ T cells, B cells, Monocytes, NK cells, NKT cells, Memory B cells, Monocyte-derived DC / Macrophages, and Platelets.
• Initial observations suggest good separation of clusters, including minor subpopulations, with Cluster 12 identified as potentially representing platelets or noise.

How to Run This Project

1. Clone the repository:

   git clone https://github.com/adabyt/scrnaseq_analysis_project.git
   cd scrnaseq_analysis_project

2. Set up the Conda environment:

   conda env create -f environment.yml
   conda activate scrnaseq_env

3. Download the dataset: Download the filtered_feature_bc_matrix.h5 file from the 10x Genomics website (link provided in the Dataset section above) and place it into a data/ subdirectory within the project root (i.e., scrnaseq_analysis_project/data/).
4. Open and run the Jupyter Notebook:

   jupyter lab

   Then, open 01_initial_data_exploration.ipynb and run all cells sequentially.

Future Work / Next Steps

As this project is for learning purposes, future explorations could include:

• Saving the final annotated AnnData object for quicker loading in future sessions.
• Performing more in-depth differential gene expression analysis between specific clusters.
• Investigating cell-cell communication using ligand-receptor analysis.
• Exploring potential cell differentiation trajectories (e.g., for the monocyte-derived populations).
• Integrating additional datasets or samples (e.g., from different conditions or donors).
• Comparing different clustering resolutions and their biological implications.