Genetic demultiplexing of single-cell and single-nuclei omics

Contents of this repository

This repository contains a Snakemake workflow that enables genetic demultiplexing of single-cell, single-nuclei, CITEseq and paired Multiome (RNA+ATAC) data.

Installation and dependencies

Conda environments

Conda YAML files are provided for all the tools in envs/. Users can create named environments from these files by running conda env create -f {file}. These environments contain most of the dependencies but not all, see below for other tools that need to in $PATH.

External dependencies

Some tools require additional dependencies, listed below. These tools can either be installed centrally and consequently symlinked (e.g. ln -s) inside the miniconda env bin directory, or installed inside each miniconda env separately.

Environment:

• Souporcell: minimap2, samtools, vartrix
• Vireo: cellsnp-lite
• Demultiplex_atac: subset-bam, samtools, bwa mem, vartrix

For the following rules, the corresponding R packages need to be installed and accessible:

• DoubletFinder: DoubletFinder
• ArchR: ArchR
• EmptyDrops: DropletUtils
• Soupx: soupx

Running the workflow

This github contains a shell script that accepts the following commands:

• ./run: Without argument, will run the normal workflow entirely based on user-given input (see below)
• ./run dry: dry run using Snakemake's -p -n arguments
• ./run unlock: Unlock snakemake directories after sigkill.

Running these commands from the top-level directory will execute the appropriate Snakemake commands.

Configuring Snakemake on a high-performance cluster or job scheduling system

It is not necessary to change the underlying code to configure this pipeline to work on a different scheduler or HPC. Instead, adjust config/config.yaml to your needs and the local installation. For many different schedulers, snakemake configs are readily available.

Input and user-specified parameters

The input of the workflow is an excel-sheet that must contain the following columns:

• pool: The name of the pool to be demultiplexed, will be name of the output folder too
• datalocation: Location of top-level Cellranger folder
• donors: nr of donors multiplexed in that specific pool
• vcf: Reference genotype file. Recommend to impute first.
• comment: May be empty, is not used.
• profile: Datatype, either RNA/ATAC/Multiome for now.

Change the name of the excel sample sheet at the top of the Snakemake file (dmx.smk), and change the output directory here too, along with the location of the CellRanger reference genomes and indices.

Author

Martijn Zoodsma, martijn.zoodsma@helmholtz-hzi.de Centre for Individualised Infection Medicine, Helmholtz Centre for Infection Research, Hanover, Germany.

Principal investigator: Yang Li Centre for Individualised Infection Medicine, Helmholtz Centre for Infection Research, Hanover, Germany.