Skip to content

Easily assess quality control data across multiple single cell datasets

License

Notifications You must be signed in to change notification settings

compugen-ltd/supercells

Repository files navigation

Supercells

Easily assess quality control data across multiple single cell datasets

Supercells is a tool which generates a single report for multiple 10x cellranger samples - thus saving time and allowing comparative QC analysis.

It is written in Python (tested with v3.7-3.9).

Code style: black PyPI Version Downloads Compugen on Twitter


📦 Installation

Using PyPI:

pip install supercells

Clone or download the repository and navigate to the package directory, then invoke:

pip install .

💻 Usage

Once supercells is installed it is run by invoking:

supercells -i <experiment folder>

Unless specified otherwise it will generate supercells_report.html within the input folder as well as supercells_report folder containing summarizing xlsx file and CSV with the raw data. To specify an output directory use the -o <output folder> flag

image

More options

usage: supercells [-h] [--version] --input INPATH [--output OUTPATH] [--cutoff-dict CUTOFF_DICT_PATH]

supercells: Easily assess quality control data across multiple single cell datasets

options:
  -h, --help            show this help message and exit
  --version, -v         show program's version number and exit
  --input INPATH, -i INPATH
                        Enter the location of the input files
  --output OUTPATH, -o OUTPATH
                        Specify the location of the output files, default is current wd
  --cutoff-dict CUTOFF_DICT_PATH, -c CUTOFF_DICT_PATH

cutoff-dict parameters is a file path to a json file. This is a sample of the json:

{
    "Mean Reads per Cell": 20000,
    "Median Genes per Cell": 1500,
    "Valid Barcodes": "75%",
    "Sequencing Saturation": "30%",
    "Q30 Bases in RNA Read": 65,
    "Fraction Reads in Cells": "70%",
    "Reads Mapped to Genome": "75%",
    "Reads Mapped Confidently to Transcriptome": "30%"
}

The keys are the field names and the values are cutoff values for that field. i.e. Mean Reads per Cell cutoff
value is 20000. So for each sample if the Mean Reads per Cell is less than 20000 the output would show
that cell in red, else in green.

Further development

Suggestions for additional features and code contributions are welcomed

About

Easily assess quality control data across multiple single cell datasets

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages