Single-cell Analysis Pipeline

Author: Geoff Dilly, PhD
ORCID: 0000-0002-6277-9932
Lab: Messing Lab, UT Austin

This repository provides a streamlined pipeline for analyzing 10x Chromium single-cell data using R and Seurat. The pipeline performs QC, clusters cells, identifies marker genes, and generates plots and CSV files in a recommended directory structure. This pipeline is intended for neuroscience research in the Messing Lab at UT Austin, and can be adapted for similar use.

• install.sh: Bash script to generate the recommended file structure in a local directory.
• setup_env.sh: Bash script to make an Anaconda environment with the appropriate dependencies.
• sc_analysis_env.yaml: YAML containing Anaconda dependencies.
• R/: R scripts for data processing, clustering, cell identification, visualization, and DGE.
• run_pipeline.sh: Editable Bash script to run the R scripts in order.
• sc_experiment_config.yaml: YAML configuration file to set pipeline parameters.
• sc_sample_metadata.csv: Metadata CSV file for identifying and labeling samples.
• reference/marker_gene_db.csv: A CSV database of sets of marker genes for cluster analysis.

1. Clone the repository:

   git clone https://github.com/Geoff-Dilly/GD_scRNA_Pipeline
   cd <repo directory>

2. Install file structure:

   bash install.sh

3. Set up Conda environment:

   bash setup_env.sh
   conda activate sc_analysis_env

4. Set up metadata and configuration:
   • Edit sc_experiment_config.yaml and sc_sample_metadata.csv as needed
5. Run the analysis pipeline:

   bash run_pipeline.sh

6. Examine outputs: View plots, results, and logs.

• raw_data/ — Raw input files (10x count matrices)
• r_data/ — RDS files of Seurat objects
• plots/ — Output figures and QC plots
• csv_results/ - CSVs of cell counts, markers, and DGE results
• logs/ — Run logs and script backups
• R/ — R scripts

• Metadata is a CSV that must contain four required columns: Sample_name, Sex, Treatment, Raw_data_dir
• Raw_data_dir should direct to the Cell Ranger outs/ folder containing filtered_feature_bc_matrix/
• Additional columns are automatically read into the Seurat object as sample-level metadata

Example:

Saved as a CSV:

Sample_name,Sex,Treatment,Age,Raw_data_dir
Subject_1,M,Drug,24,"raw_data/Subject_1/outs"
Subject_2,F,Ctrl,27,"raw_data/Subject_2/outs"
Subject_3,M,Ctrl,27,"raw_data/Subject_3/outs"
Subject_4,F,Drug,26,"raw_data/Subject_4/outs"

• Marker genes can be stored in a CSV database in references/marker_gene_db.csv
• A reference from this database will be used for cell-identification plots
• Default marker genes (major brain cell types) come from Dilly et al. (2022)
• Custom references can be added
• The reference that will be plotted can be set with scConfig$marker_gene_reference

Example:

• R = 4.3.3
• Seurat ≥ 5.0
• Anaconda
• See sc_analysis_env.yaml for dependencies

This repository is maintained for experiments in the Messing Lab at UT Austin.