Morphmap

Table of contents

• Metadata
• Images
  • How to download images?
• Extracting single-cell features using CellProfiler
  • Creation of single-cell profiles
    • How to download single-cell profiles?
  • Creation of well-level profiles
    • How to download well-level profiles?
• Processing the profiles
• How to run the analyses?
  • Cloning this repo
  • Downloading the profiles
  • Running the analyses notebooks
    • Installing the conda environment

Table of contents generated with markdown-toc

Link to the biorxiv manuscript: Not yet available

This is a dataset of images and profiles generated as a part of the JUMP Cell Painting (JUMP-CP) project. Genes were either over-expressed (ORF) or knocked out (CRISPR) and the cells were assayed using an imaging assay called Cell Painting. From the images, features were extracted using the CellProfiler software. The features were then processed and the resulting profiles were the analyzed using notebooks in this repository.

In the following sections, instructions are provided for downloading the various components of this dataset, processing the dataset and analyzing the profiles.

Metadata

Metadata information, such as, which plate from which batch contains a particular gene, is available in the datasets repo.

Images

Cell images are available for download from the cellpainting gallery public AWS S3 bucket.

How to download images?

There are two sources of data. ORF images are from source_4 and CRISPR images are from source_13.

source=<SOURCE NAME>
aws s3 sync \
  --no-sign-request \
  s3://cellpainting-gallery/cpg0016-jump/${source}/images/ . 

Extracting single-cell features using CellProfiler

Features were extracted using the CellProfiler pipeline in https://github.com/broadinstitute/imaging-platform-pipelines/tree/master/JUMP_production#production-pipelines.

Creation of single-cell profiles

Instructions for creating the single-cell profiles from images are provided in the Image-based profiling handbook.

How to download single-cell profiles?

Single-cell profiles can be downloaded from the cellpainting gallery public AWS S3 bucket.

source=<SOURCE NAME>
batch=<BATCH NAME>
plate=<PLATE NAME>
aws s3 sync \
  --no-sign-request \
  s3://cellpainting-gallery/cpg0016-jump/${source}/workspace/backend/${batch}/${plate}/ --exclude "*" --include "*.sqlite" .

Creation of well-level profiles

Well-level profiles are also created using the instructions provided in the Image-based profiling handbook.

How to download well-level profiles?

Well-level profiles can also be downloaded from the cellpainting gallery public AWS S3 bucket.

source=<SOURCE NAME>
batch=<BATCH NAME>
plate=<PLATE NAME>
aws s3 sync \
  --no-sign-request \
  s3://cellpainting-gallery/cpg0016-jump/${source}/workspace/profiles/${batch}/${plate}/ .

Processing the profiles

Various steps were performed to remove technical noise from the profiles. These steps are as follows:

• Well position correction
• Cell count regression
• Normalization
• Outlier removal
• Feature selection
• Sphering
• Harmony correction

These steps can be performed using the jump-profiling-recipe and the appropriate config file (orf.json and crispr.json from the input folder).

The processed profiles are stored in the cellpainting-gallery bucket.

How to run the analyses?

Cloning this repo

To download/clone this repository, run the following commands

git clone https://github.com/jump-cellpainting/2024_Chandrasekaran_Morphmap.git 
cd 2024_Chandrasekaran_Morphmap
git submodule update --init --recursive

Downloading the profiles

To download the profiles and other files required to run the analyses in this repository, run the following commands

cd profiles
./download-profiles.sh

Running the analyses notebooks

Notebooks in each folder are run to replicate the analyses. To reproduce the analyses, install the conda environment within each folder and run the notebooks.

Installing the conda environment

Download and install mamba from miniforge using the appropriate installer for your operating system.

Once installed, run the following command to create the conda environment in each folder using the following commands

mamba env create -f environment.yml
mamba activate <conda environment name>