Cellranger workflow

Workflow for processing sc-multiome and sc-atacseq data using CellRanger

Dependencies

Dependencies	Link
CellRangerARC	https://www.10xgenomics.com/support/software/cell-ranger-arc/latest
CellRangerATAC	https://support.10xgenomics.com/single-cell-atac/software/downloads/latest
SRA_Toolkit	https://github.com/ncbi/sra-tools/wiki/01.-Downloading-SRA-Toolkit
Snakemake	https://snakemake.readthedocs.io/en/stable/
PigZ	https://zlib.net/pigz/

Debugging & Running

To Debug

snakemake -p --dry-run -s <insert snakefile name>

To Run

snakemake --cores all --keep-incomplete --resources load=100

Resources flag must be added when running Snakemake pipeline to ensure proper load management; modify according to computing power.

[For reference, 1 run_cr requires a load of 50; 256GB of ram allows for ~ 2 simultaneous runs of CellRanger.]

Workflow

#1 Preparation of "metadata.csv"

Most important step of entire workflow. In this example, the file test_run_multiome.csv is used in the config.yaml file - any file name can be used just edit config file accordingly. Ensure format of xxx_metadata.csv is adhered to strictly; Format should be exactly as that of the "xxx_metadata.csv" file provided.

Note: "GSM" typically refers to a library and "SRR" to a run. Although these terminologies may be used inaccurately in the current context, please adhere to the column names specified in test_run_multiome.csv for consistency and accuracy.

#2 Adjusting parameters in config.yaml

The config.yaml file looks as follows:

#ATAC or multiome
type: "multiome"

#true or false
remove_fastq: 

path:
  #absolute path to working directory
  wd: "/home/users/astar/gis/stufrancis/scratch/RETINA_ATLAS/test_run_multiome"
  
  #relative path to reference genome, relative to snakefile
  ref_genome: "../../CellRangerFiles/refdata-cellranger-arc-GRCh38-2020-A-2.0.0"
  
  #absolute path to metadata.csv file generated in previous step
  metadata: "/home/users/astar/gis/stufrancis/scratch/RETINA_ATLAS/test_run_multiome/01_prep_metadata/test_run_multiome.csv"
  
  #absolute path to "02_snakemake_download_cellranger" process directory
  process_dir: "/home/users/astar/gis/stufrancis/scratch/RETINA_ATLAS/test_run_multiome/02_snakemake_download_cellranger"

#suffix of reads from SRA toolkit to be renamed to
#if the read does not exist, substitute with "ignore"
ATAC1 : "S1_L001_R1_001.fastq"
ATAC2 : "S1_L001_R2_001.fastq"
ATAC3 : "S1_L001_R3_001.fastq"
ATAC4 : "ignore"

RNA1 : "S1_L001_R1_001.fastq"
RNA2 : "S1_L001_R2_001.fastq"
RNA3 : "ignore"
RNA4 : "ignore"

##the name of the files must end in .fastq

Ensure config.yaml files is properly adjusted. For more information and to know what suffix to use, please refer to: https://www.10xgenomics.com/support/software/cell-ranger-arc/latest/tutorials/inputs/specifying-input-fastq-count

Common mistakes include forgetting "ignore" in the suffixes section and mistyping respective paths, which may lead to errors.

#3 Maintaining proper directory structure & scheduling jobs

Prior to running the snakemake pipeline, ensure all files present.

```
.
├── 01_prep_metadata/
│   └── metadata.csv  
├── 02_snakemake_download_cellranger/ 
├── 03_Seurat_object/ 
├── snakemake
├── config.yaml
└── xxx.sbatch #for job scheduling on hpc
```

Runs will be generated according to "ID" in metadata.csv [i.e. 1 "ID" will have 1 "outs" folder].

Example of xxx.sbatch format to run pipeline:

#!/bin/bash

#SBATCH -t 72:00:00
#SBATCH -N 1
#SBATCH --ntasks-per-node 60
#SBATCH --cpus-per-task 1
#SBATCH --mem 128G
#SBATCH -J CR_pipeline
#SBATCH --output=%x-%j.out
#SBATCH --error=%x-%j.err
#SBATCH -p cpu

snakemake --cores all --keep-incomplete --resources load=100

## sbatch run_nfcore_scrnaseq.sbatch
## squeue --job 1201585

Snakemake rule pipeline

Graph above shows a pipeline run for "id1" and "id2"

#4 What to do after CellRanger is complete?

After the CellRanger run is complete, the "key files" for downstream analysis need to be prefixed and suffixed (extracted) and merged into one Seurat object. First, have a look at 2_merge_peaks for file extraction, and then retina_atlas.ipynb for downstream work.