Neurovascular Atlas of Alzheimer's Disease

Overview

This repository contains the analysis code for a single-nucleus RNA sequencing study of the neurovascular unit (NVU) in Alzheimer's disease (AD). We profiled prefrontal cortex samples from AD patients and age-matched controls to generate a high-resolution atlas of vascular and parenchymal cell types.

Associated manuscript: In preparation

Processing of data

In brief: standard processing was done with cellranger, using an updated reference with Ensembl version 109 and Gencode version 43. The output of this was run through scFlow with standard parameters which was then read into R via the Seurat package.

Looking at the read counts and number of features output from scFlow revealed that 3 of the donors, and the vascular fraction from a fourth donor, were of poor quality and so they were excluded from downstream analysis.

Data Availability

The raw sequencing data (10X Chromium single-nucleus RNA-seq) and processed count matrices are available from GEO, accessions GSE310554 & GSE222007. Note that GSE310554 contains the processed data of all samples from both datasets except for those donors mentioned above, and is a Seurat object.

Computational environment

There's a few options if you want to replicate the R environment used for this analysis.

Option 1: Using Docker (Recommended)

The easiest way to reproduce our analysis environment is using Docker.

Pull from Docker Hub

The image is on Docker Hub here

# Pull the image from Docker Hub
docker pull multitude5286/ad-bbb-analysis

Run the Docker container

Having got the Docker image ready you can either enter it as an interactive R session

# Interactive R session
docker run --rm -ti ad-bbb-analysis R

Or use it via RStudio like so

# Use RStudio Server (optional)
docker run --rm -ti \
  -e USER=rstudio \
  -e PASSWORD=yourpassword \
  -p 8787:8787 \
  ad-bbb-analysis \
  /init

Having run this you can visit http://localhost:8787/ and use username: rstudio, password: yourpassword

Build the Docker image yourself

Alternatively you can build it yourself with the Dockerfile in this repo

# Clone this repository
git clone https://github.com/UKDRI/AD-Neurovascular-Atlas.git
cd ad-neurovascular-atlas

# Build the Docker image (takes ~30-60 minutes)
bash build-docker.sh

# Or build manually
docker build -t ad-bbb-analysis:1.0.0 .

Option 2: Using renv (Local Installation)

If you prefer to install packages locally:

# Clone repository
git clone https://github.com/UKDRI/AD-Neurovascular-Atlas.git
cd ad-neurovascular-atlas

# Install renv if you don't have it
R -e "install.packages('renv')"

# Restore the R environment (this will take a while)
R -e "renv::restore()"

Note: You'll need to install system dependencies manually. See Dockerfile for the required system libraries.