10X Genomics scRNA-seq Pipeline: Preprocessing, Analysis & Visualization

This repository contains a modular pipeline for preprocessing, analyzing, and visualizing 10X Genomics single-cell RNA-seq (scRNA-seq) data. Developed for a Bachelor thesis at ZHAW in collaboration with Nexco Analytics, the pipeline supports alignment with STARsolo and downstream analysis using Seurat.

Repository Structure

├── Code/
│   ├── submit_pipeline.sh      # SLURM job submission wrapper
│   ├── submit_pipeline_2.sh    # Alternative SLURM logic (optional)
│   ├── run_pipeline.sh         # Core alignment and quantification script using STARsolo
│   ├── run_pipeline_2.sh       # Variant version for brain tumor dataset (optional)
│   ├── seurat_analysis.R       # R script for downstream analysis using Seurat
│   └── README.md               # Documentation explaining SLURMs and R script components
│
├── requirements/
│   └── seuratenv_enhanced.yml  # Conda environment for R
│
└── README.md

Requirements

• STAR (v2.7.10a or newer)
• R (v4.2.0 or newer)
• Conda (Anaconda or Miniconda)

Dependencies and Setup

Specific dependencies used in this project are all found in the .yml file in the requirements folder of this repository, the most important packages are listed below.

Core Analysis

• Seurat - Main single-cell analysis framework
• optparse - Command line argument parsing
• dplyr - Data manipulation
• future - Parallel processing

Visualization

• ggplot2 - Static plotting
• patchwork - Plot composition
• plotly - Interactive visualizations
• viridis - Color scales
• htmlwidgets - Web widget creation

Cell Type Annotation

• SingleR - Reference-based cell type annotation

Data Export

• jsonlite - JSON export functionality
• reshape2 - Data reshaping

To install the R environment:

conda env create -f requirements/seuratenv_enhanced.yml
conda activate seurat_env_3

Usage

To run the full pipeline on an HPC cluster:

sbatch Code/submit_pipeline.sh

This script:

• Calls run_pipeline.sh for alignment and count matrix generation using STARsolo
• Automatically creates a timestamped output folder
• Triggers seurat_analysis.R for QC, clustering, cell type annotation, and visualization

All paths and parameters can be configured inside run_pipeline.sh.

Input Datasets and Output Results

Due to GitHub file size limits, raw input data and pipeline results are hosted on OneDrive.

Download full datasets and outputs from:

https://1drv.ms/f/c/7198abcaecb08200/Eng_pHWTSZpFspANuoRADycBesz-RlqCJf-0ibVhL080Eg?e=J5oSRu

Directory structure on OneDrive:

Datasets/
├── pbmc_1k_v3_fastqs/          # PBMC 1k dataset (2 sequencing lanes)
├── Brain_Tumor_3p_fastqs/      # Brain tumor dataset (4 sequencing lanes)

Outputs_Pipeline/
├── Output_1k_PBMC/             # STARsolo + Seurat output for PBMC
├── Output_brain_tumor_3p/      # STARsolo + Seurat output for brain tumor

Output Overview

Each pipeline run produces:

• STARsolo output (barcodes, genes, matrix.mtx)
• Filtered cell matrices and statistics
• Seurat-based downstream results:
  • QC metrics and violin plots
  • UMAP/t-SNE dimensionality reductions
  • Cell clustering and annotations
  • Differential expression and functional scoring
  • Interactive Dashboard

A more detailed documentation about the results can be found on the OneDrive folder

Base Directory Structure used in the project

/cfs/earth/scratch/paleslui/BATH/
├── genomeDir/                      # STAR genome index
│   ├── Genome                      # Genome sequences
│   ├── SA                          # Suffix array
│   ├── SAindex                     # Index files
│   └── whitelist.txt              # Valid cell barcodes
├── pbmc_1k_v3_fastqs/             # PBMC dataset
│   ├── pbmc_1k_v3_S1_L001_R1_001.fastq.gz
│   └── pbmc_1k_v3_S1_L001_R2_001.fastq.gz
├── Brain_Tumor_3p_fastqs/         # Brain tumor dataset
│   ├── Brain_Tumor_3p_S2_L001_R1_001.fastq.gz
│   ├── Brain_Tumor_3p_S2_L001_R2_001.fastq.gz
│   ├── ... (L002, L003, L004)
├── R/                             # R analysis scripts
│   └── seurat_analysis_complete_12.R
└── SLURMs/                        # Pipeline scripts
    ├── submit_pipeline.sh
    ├── submit_pipeline_2.sh
    ├── run_pipeline.sh
    └── run_pipeline_2.sh

Credits

This pipeline was developed by Luigi Palese as part of a Bachelor thesis in Applied Digital Life Sciences at ZHAW. Project supported by Nexco Analytics.