Skip to contents

Workshop website: https://phipsonlab.github.io/single_cell_workshop/

Overview

Single-cell RNA sequencing (scRNA-seq) has revolutionised our ability to study gene expression at the resolution of individual cells, enabling detailed characterisation of cell types and their association with complex diseases and phenotypes. This workshop provides a comprehensive introduction to the computational analysis of scRNA-seq data using R and Bioconductor, including some advanced topics such as continuous cell state modelling and co-expression network analysis.

We analyse single-nucleus RNA-sequencing (snRNA-seq) data from human heart tissue across three developmental stages: foetal, young, and adult. The dataset originates from Sim et al. (2021) examining sex-specific control of human heart maturation (Circulation).

Pre-requisites

This workshop is designed for researchers and students who:

  • Have basic familiarity with R programming (data manipulation, plotting)
  • Are interested in single-cell transcriptomics analysis
  • Want to understand best practices for scRNA-seq data processing

Some knowledge of single-cell analysis is recommended. All concepts are introduced from first principles with detailed explanations.

System Requirements

Resource Minimum Recommended
RAM 8 GB 16 GB
Disk space 5 GB free 10 GB free
R version 4.3+ 4.5.2
RStudio 2023.06+ Latest

Supported platforms: Windows 10/11, macOS 12+ (Intel and Apple Silicon), Ubuntu 22.04+ / equivalent Linux. A C/C++ build toolchain is required on each platform — Module 0 walks through the install (Rtools45 on Windows, Xcode Command Line Tools on macOS, build-essential + dev headers on Linux).

Workshop Outline

Session 1: Core Single Cell Analysis (Morning, ~3 hours)

Module Topic Duration
Module 1 Quality Control 30 min
Module 2 Normalisation & Clustering 50 min
Module 3 Cell Type Annotation 20 min
Module 4 Differential Expression 50 min
Wrap-up & Q&A 10 min

Session 2: Continuous Cell States (Afternoon, ~3 hours)

Module Topic Duration
Module 5 Continuous Phenotyping with Φ-Space 45 min
Module 6 Pseudotime Trajectory Analysis 45 min
Module 7 Cell-specific Co-expression Networks (NeighbourNet) 45 min

Learning Objectives

By the end of this workshop, participants will be able to:

  • Load and explore 10X Genomics scRNA-seq data in R using Seurat
  • Calculate and interpret per-cell quality control metrics
  • Apply appropriate filtering thresholds to remove low-quality cells
  • Normalise data using SCTransform and correct batch effects with Harmony
  • Perform graph-based clustering and visualise results with UMAP
  • Annotate cell types using canonical marker genes
  • Understand the pseudoreplication problem in single-cell differential expression
  • Perform statistically rigorous differential expression analysis using pseudobulk methods
  • Analyse cell type composition changes using propeller
  • Replace hard cell-type labels with continuous Φ-Space phenotype scores
  • Fit and compare pseudotime trajectories (slingshot, DPT) on multiple embeddings
  • Build cell-specific co-expression meta-networks with NeighbourNet

Dataset

The workshop uses snRNA-seq data from human heart tissue (Sim et al., 2021):

Group Samples Age Range Description
Foetal 3 19-20 weeks Developing heart
Young 3 4-14 years Postnatal maturation
Adult 3 35-42 years Mature heart

Total: 9 samples, ~47,000 nuclei after quality control

Methods Covered

Analysis Step Method Package
Quality control Per-cell metrics, filtering Seurat
Normalisation SCTransform v2 Seurat, glmGamPoi
Batch correction Harmony harmony
Dimensionality reduction PCA, UMAP Seurat
Clustering Louvain algorithm Seurat
Cell type annotation Marker-based (manual) Seurat
Differential expression Pseudobulk + limma-voom edgeR, limma
Composition analysis propeller speckle
Soft annotation PLS on reference atlas PhiSpace
Pseudotime Principal curves + diffusion pseudotime slingshot, destiny
Co-expression networks Cell-specific networks + meta-networks NeighbourNet

Quick Start

Please complete setup at least one day before the workshop.

  1. Clone or download this repository (Windows users: clone to a short path like C:\workshop\ to avoid the 260-character path limit).
  2. Open single_cell_workshop.Rproj in RStudio.
  3. Follow Module 0: Setup from start to finish.

The setup runs as a single unified flow that covers both sessions:

  • Step 2 — System build tools (Rtools45 on Windows, Xcode CLT on macOS, build-essential on Linux). Required because a few packages compile from source.
  • Step 3 — R packages: renv::restore() for the locked core, then BiocManager::install(...) + remotes::install_github(...) for the extras (PhiSpace, NeighbourNet, slingshot, destiny, scater, ComplexHeatmap).
  • Step 4 — Workshop data from Zenodo (~420 MB).

Total time: roughly 20–40 minutes depending on whether the GitHub-only packages need to compile from source.

Optional: Backup checkpoints

A separate Zenodo record hosts pre-computed checkpoints so you can start at any module boundary — useful for skipping straight to a particular technique, or for starting Session 2 without first running Session 1. Each file replaces the output of one or more upstream modules:

File Lets you skip
01_qc_filtered.rds Module 1
02_integrated_clustered.rds Modules 1 + 2
03_annotated.rds Modules 1 + 2 + 3
afternoonSession.zip All of Session 1 — start at Module 5

afternoonSession.zip contains the Session 2 (Module 5, 6, 7) input and intermediate results. Download, unzip, and the files land in data/ and results/ per the instructions in Module 0.

The download chunk lives in Module 0, “Optional: Backup Checkpoints”.

Key Package Versions

The core packages are pinned in renv.lock for reproducibility. The afternoon-session extras are installed at the latest Bioconductor 3.22 / GitHub HEAD versions (see Module 0 Step 3b).

Package Source Package Source
R 4.5.2 renv.lock Bioconductor 3.22 renv.lock
Seurat 5.4.0 renv.lock edgeR 4.8.2 renv.lock
SeuratObject 5.3.0 renv.lock limma 3.66.0 renv.lock
harmony 1.2.4 renv.lock speckle 1.10.0 renv.lock
glmGamPoi 1.22.0 renv.lock
ComplexHeatmap Bioconductor slingshot Bioconductor
destiny Bioconductor scater Bioconductor
PhiSpace GitHub (jiadongm/PhiSpace) NeighbourNet GitHub (meiosis97/NeighbourNet)

Workshop Materials

Session 1: Core Single Cell Analysis

Module Topic Description
Module 0 Setup Environment setup and data download
Module 1 Quality Control QC metrics, cell filtering
Module 2 Integration Normalisation, batch correction, clustering
Module 3 Annotation Marker genes and cell type assignment
Module 4 DE Analysis Pseudobulk DE and composition analysis

Session 2: Trajectory and Gene Regulation

Module Topic Description
Module 5 Continuous Phenotyping with Φ-Space Soft cell-type + stage scores via PLS on a reference atlas
Module 6 Pseudotime Trajectory Analysis Slingshot and DPT on PCA and Φ-Space embeddings
Module 7 Cell-specific Co-expression Networks NeighbourNet meta-networks from maturation-associated targets

Citation

If you use materials from this workshop, please cite:

Original dataset:

Sim CB, Phipson B, Ziemann M, et al. Sex-Specific Control of Human Heart Maturation by the Progesterone Receptor. Circulation. 2021;143(10):1614-1628. doi:10.1161/CIRCULATIONAHA.120.051921

Acknowledgements

This workshop was developed by the Phipson Lab using data from the Porrello and Hewitt laboratories. We thank the original authors for making their data publicly available.

License

This project is licensed under the MIT License - see the LICENSE file for details.