scRNAseq-repo

This repository contains the code for the main analyses performed in the "Comparative Analysis of Single Cell RNA Sequencing Technologies" paper (De Simone and Hoover, et. al).

Main editors for this repo are Jonathan Hoover and Daniel Le. For correspondence regarding code please contact Daniel Le ([email protected]). For correspondence regarding experimental design please contact Spyros Darmanis ([email protected]).

scRNA kits processed in this project include:

• Chromium Single Cell 3' Reagent Kit, v3.1 Chemistry. (10x Genomics): '10X_3'
• Chromium Single Cell 5' Reagent Kit, v2 Chemistry. (10x Genomics): '10X_5'
• Chromium Fixed RNA Profiling Reagent Kit (10x Genomics): '10X_FRP'
• PIPseq T20 3' Single Cell RNA Kit v4.0 (Fluent Biosciences): 'Fluent'
• BD Rhapsody WTA Reagent Kit (Becton Dickinson): 'BD'
• HIVE CLX scRNA Seq Kit v1 (Honeycomb Biotechnologies): 'Honeycomb'
• Evercode WT v2 (Parse Biosciences): 'Parse'
• Single Cell RNA Kit (Scale Biosciences): 'Scale'
• ASTERIA Single-cell RNASeq Kit (Scipio Bioscience): 'Scipio'

Reproducibility

Conda Environments

Analyses were performed using the mamba environment shown in ./mamba_environments/scanpy-default-mamba.yml. To install this virtual environment using miniforge, make sure miniforge is installed (see https://github.com/conda-forge/miniforge) and then execute the following:

First, clone this directory to your desired local directory (make sure git is installed)

cd </path/to/clone/location>
git clone https://github.com/danledinh/scRNAseq-repo.git

Then, cd into the cloned directory and create the conda repo from the .yml file provided with your desired

cd </path/to/scRNAseq-repo>
conda env create -f ./mamba_environments/scanpy-default-mamba.yml -n <envname>

Annotated AnnData Objects

After setting up the conda environment, some notebooks can be be run directly using data deposited on the CZ Biohub's CellxGene repo (see the Resources section below). Some just act as reference for how data was generated. Each notebook will mention whether it can be rerun directly using the resources made available within this repo.

For those that are able to be run using the AnnData objects uploaded to CellxGene, please run the notebooks/CellxGene_Download.ipynb notebook first. It will generate a results/anndata_objects/ directory containing all the annotated .h5ad files for each kit along with the harmony integrated .h5ad file.

scanpy-compatible Broad Reference Dataset

The raw dataset used to train annotation models (CellTypist and Seurat's Label Transfer) can be downloaded and converted to a .h5ad file with the notebooks/BroadInstituteReference_convertRDStoH5.Rmd file. This reference dataset is hosted in the SeuratData package (https://github.com/satijalab/seurat-data) and is from a 2020 comparative analysis of non-commercial single-cell RNA sequencing technologies on PBMCs (https://www.nature.com/articles/s41587-020-0465-8).

The data was processed and subsampled for use and training according to the notebooks/analysis/ProcessandFilterCells.ipynb notebook. It was then used to train two separate annotation models as described in our paper (see the Resources section below).

Resources

The following are links to relevant resources related to this project:

• Pre-print: https://www.biorxiv.org/content/10.1101/2024.06.18.599579v1

• SRA BioProject: PRJNA1106903 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1106903)

• CZ Biohub CellxGene Collection: 398e34a9-8736-4b27-a9a7-31a47a67f446 (https://cellxgene.cziscience.com/collections/398e34a9-8736-4b27-a9a7-31a47a67f446)