campbio/sctk_qc repository overview

⁠Generation of comprehensive quality control metrics with SCTK

This pipeline will import data from single-cell preprocessing algorithms (e.g. CellRanger), generate various quality control metrics (e.g. doublet scores), and output results in standard data containers (e.g. SingleCellExperiment). Both the original droplet matrix and the filtered cell matrix will be processed. This pipeline is focused on single cell data generated from microfluidic devices (e.g. 10X).

⁠Docker

If you have not used docker before, you can follow the instruction to install and set up docker in Windows⁠, Mac⁠ or Linux⁠.

The Docker image can be obtained by running:

docker pull campbio/sctk_qc

Noted that the transcriptome data and GMT file needed to be accessible to the container via mounted volume. In the below example, mount volumn is enabled for accessing input and output directory using argument -v. To learn more about mounted volumes, please check out this post⁠.

The usage of each argument is the same as running command line analysis. Here is an example code to perform quanlity control of CellRangerV3 data singleCellTK docker:

docker run --rm -v /path/to/data:/SCTK_docker \
-it campbio/sctk_qc:latest \
-b /SCTK_docker/cellranger \
-P CellRangerV3 \
-s pbmc_100x100 \
-o /SCTK_docker/result/tenx_v3_pbmc \
-g /SCTK_docker/mitochondrial_human_symbol.gmt \
-S TRUE \
-F R,Python,Flatfile,HTAN

⁠Singularity

The Singulatiry image can easily be built using Docker Hub as a source:

singularity pull docker://campbio/sctk_qc:latest

The usage of singleCellTK Singularity image is very similar to that of Docker. In Singularity 3.0+, the mount volume is automatically overlaid⁠.

It's recommended to re-set the home directory when you run singularity. Singularity will mount $HOME path on your machine by default, which might contain your personal R/Python library folder. If we don't re-set the home to mount, singularity will try to use R/Python libraries which are not built within the singularity image and cause some conflicts. You can point to some "sanitized home", which is different from $HOME path on your machine, using argument -H/--home⁠. Besides, you can use argument --bind/-B to specify your own mount volume, which is the path that contains the dataset and will be used to store the output of QC pipeline. The example is shown as below:

singularity run --home=/PathToSanitizedHome \
--bind /PathToData:/data sctk_qc_2.12.1.sif \
-P CellRangerV3 \
-s gencodev34_pbmc_1k_v3 \
-b /data/gencodev34_pbmc_1k_v3
-o /data/result/gencodev34_pbmc_1k_v3 \
-S TRUE \
-F R,Python,FlatFile,HTAN \
-n 15 \
-T MulticoreParam

Note that you may need to update the version number from "2.12.1" to whatever version you are using. Also, you might want to specify cpu architecture when run the Singularity image on BU SCC using '#$ -l cpu_arch=broadwell|haswell|skylake|cascadelake' command. Because the python packages are compiled by SIMD instructions that are only compatible on these cpu architectures. If you are runnning singularity on other cluster, please contact IT helps about how to specify cpu architecture when you run the singularity image. One of the example is shown below:

#!/bin/bash
#$ -cwd
#$ -j y
#$ -P camplab
#$ -pe omp 16
#$ -l cpu_arch=broadwell|haswell|skylake|cascadelake
singularity run --home=/PathToSanitizedHome \
--bind /PathToData:/data sctk_qc_2.12.1.sif \
-P CellRangerV3 \
-s gencodev34_pbmc_1k_v3 \
-b /data/gencodev34_pbmc_1k_v3
-o /data/result/gencodev34_pbmc_1k_v3 \
-S TRUE \
-F R,Python,FlatFile,HTAN \
-n 15 \
-T MulticoreParam

⁠Detailed manual

For detailed information about the parameters of the pipeline, please go to the documentation on https://camplab.net/sctk⁠, Click "Latest Documentation", and select the "SCTK-QC" tab.