G-quadruplex Profiling in Complex Tissues Using Single-cell CUT&Tag

This repository contains the main scripts for the manuscript "G-quadruplex Profiling in Complex Tissues Using Single-cell CUT&Tag" as well as the processed Seurat objects used in the analysis. Main Steps Include:

• Seurat processing of G4 scCUT&Tag data
• Exploratory analysis and performance validation
• Integration with scRNA-Seq data (Bartosovic et al.)
• Differential G4 analysis
• Cicero exploration

Repository Structure

• scripts/data_processing/: Contains R scripts for processing raw data (CellRanger output).
• scripts/data_integration/: Includes workflows for data integration using Seurat or scBridge.
• scripts/evaluation/: Contains scripts for exploratory analysis and downstream steps.
• utils/: Provides helper functions used during the analysis.

Workflow:

mESC-MEF G4 scCUT&Tag schematic workflow:

mESC-MEF Seurat object:

• results/Seurat/mESC_MEF/outputs/Seurat_object.Rds

postnatal mouse brain G4 scCUT&Tag schematic workflow:

unsorted Seurat object:

• results/Seurat/unsorted_mousebrain/res0.1/outputs/Seurat_object.Rds

GFP sorted Seurat object:

• results/Seurat/GFP_sorted_mousebrain/res0.8/outputs/Seurat_object.Rds

scBridge integration

scBridge is a neural network driven single-cell multi-omics data integration tool taking advantages of the existing data heterogeneity (Yunfan Li et al., 2023)

Installation: scBridge github

Note: scBridge runs on a single GPU.

1. create_h5ad.py - Create h5ad format from gene activity scores and scRNA-Seq counts
2. scbridge.sh - Run on cluster with GPU
3. scbridge_outputs.py - Visualization

Data

Seurat objects of the repository are managed with Git LFS and users will need to install it to retrieve the file.

GEO repository: