nasiegel88 / 10x-snake

General snakemake pipeline to generate cell matrices for single-cell analysis and .loom files to assess cell development

10x-snake

Author: Noah Siegel

Email: [email protected]

Last update: 2021-03-30

General snakemake pipeline to generate cell matrices for single-cell analysis and .loom files to assess cell development.

Summary

• Usage
• Output
• Next Steps
• References

Usage:

I. Anaconda

Anaconda is a requirement as it manages the packages in the workflow and can be installed here.

Clone this repo from Github on a computer that can allocate a minimum of 60Gb of RAM (most likely a cluster).

git clone https://github.com/nasiegel88/10x-snake.git
cd 10x-snake
# setup snakemake
conda env create --name snake --file env/snake.yaml
# activate environment
conda activate snake

II. Cellranger

Go to the cellranger download page and install the latest version of cellranger. You will need to enter you name, email, and institution.

https://support.10xgenomics.com/single-cell-gene-expression/software/downloads/latest

Once you have the link to download the lastest installation of cellranger replace the url in ranger-build.sh:

mkdir apps
cd apps

# New url here
curl -o cellranger-5.0.1.tar.gz "https://cf.10xgenomics.com/releases/cell-exp/cellranger-5.0.1.tar.gz?Expires=1613392910&Policy=eyJTdGF0ZW1lbnQiOlt7IlJlc291cmNlIjoiaHR0cHM6Ly9jZi4xMHhnZW5vbWljcy5jb20vcmVsZWFzZXMvY2VsbC1leHAvY2VsbHJhbmdlci01LjAuMS50YXIuZ3oiLCJDb25kaXRpb24iOnsiRGF0ZUxlc3NUaGFuIjp7IkFXUzpFcG9jaFRpbWUiOjE2MTMzOTI5MTB9fX1dfQ__&Signature=jCzQaXWhnVZFXxqfIeJCD0gnz0ULZoHPkntqQ-gIiMu~iuUWONLycXrxB4U7QpuCX9Z62~UTfGVdQM-T90NgEm6yg4Krcys5vys-FAzK48vKLima7xB8efUOd4W~sZ0YAnKU1lve4qlLI79Hd81tDHPQr162-2IakmvUxpW4GpM6tZzCr1FUxoWEFz8Z-mS3Ixl7nhB22i3bR6KngA1pR8I1S-pkwSGrltsm9OHgDVENAR25CSBVPqEqyZrjAuIclDAZzI6p40GgRIYM5z3NpRb3cTyQjl9B7CogA1DqvrIYSwChKfnh2j6O6cBZanav30K3ktHYTwm-DGcGAUvMtw__&Key-Pair-Id=APKAI7S6A5RYOXBWRPDA"

tar -xzvf cellranger-5.0.1.tar.gz

Next, run the following command.

bash ranger-build.sh

Note, you should only have to do the above for the first time running 10x-snake. If cellranger is already installed on your system you will still need to perform the above to add cellranger to the path of this directory.

III. Cluster

Edit the configuration file to point the to path of your 10x reads and confirm everything is setup by doing a dry run.

config.yaml:

proj_name: name
name: optional
email: optional

# Directory
indir: /path/to/10x/reads

# Fastq file suffix, following the read pair designation
suffix: _001.fastq.gz

# Read pair designations
r1_suf: R1
r2_suf: R2

# Parameters
cell_number: 10000
threads: 60

Masking repeats

If there is an issue with rule mask_off, download a fresh repeat masking file since there is currently no way to wget or curl the annotation file from the UCSC Genome Browswer. In order to include a download link of the annotation file in the Snakemake file the masking output file must be sent to Galaxy in order to construct a temporary symbolic link. This link can substituted into rule mask_off in the Snakefile:

rule mask_off:
    output: "repeat_msk.gtf"
    # download mouse repeat annotation gtf
    shell: 
        """
        curl -L [repeat gtf link] \
        -o {output}
        """

The current annotation GTF being used is from the Dec. 2011 (GRCm38/mm10) assembly as 10x is using this assembly for their reference database as of 2021-03-30. The annotation file can be download here

Dry run:

snakemake -np --use-conda

If there are no errors submit the job to a cluster to run the analysis. This pipeline assumes Farm is being used however, the submit script can be adjusted to run on any computer cluster.

sbatch submit.sh

Outputs:

10x-snake
	     ├── apps
	     │   ├── cellranger-5.0.1
	     │   └── cellranger-5.0.1.tar.gz
	     ├── config.yaml
	     ├── env
	     │   ├── snake.yaml
	     │   └── velocyto.yaml
	     ├── output
	     │   ├── ETS_NEO_M1
	     │   ├── FA_NEO_MI
	     │   └── logs
	     ├── ranger-build.sh
	     ├── README.md
	     ├── ref
	     │   └── refdata-gex-mm10-2020-A
	     ├── repeat_msk.gtf
	     ├── Snakefile
	     ├── stamps
	     │   ├── count_stamps
	     │   └── vel_count_stamps
	     └── submit.sh

• Note that there is no limit to the number of samples that can run through 10-snake.

Next Steps...

Most of the single analysis pipeline can be done in R using the package called Seurat. Here are some useful vignettes to get started with the downsteam analysis:

• Normalization: scTransform normalization
• Integration of multiple samples: Integration using scTransform
• Comparing treatement groups: Fast integration
• Differential expression analysis: DEG
• RNA velosity: Estimating RNA Velocity using Seurat and scVelo

References:

1. Hafemeister et al., Genome Biology 2019
2. Stuart, Butler, et al., Cell 2019
3. Butler et al., Nat Biotechnol 2018
4. La Manno et al, Nature 2018