This repository contains all supporting analyses and files for Ramezani, Bauman, Singh, and Weisbart, et al. "A genome-wide atlas of human cell morphology". It also includes download information for the images and profiles used in the analysis. It is currently under review for publication. It is available as a preprint on bioRxiv.
Meraj Ramezani, Julia Bauman, Avtar Singh, Erin Weisbart, John Yong, Maria Lozada, Greg P. Way, Sanam L. Kavari, Celeste Diaz, Marzieh Haghighi, Thiago M. Batista, Joaquín Pérez-Schindler, Melina Claussnitzer, Shantanu Singh, Beth A. Cimini, Paul C. Blainey‡, Anne E. Carpenter‡, Calvin H. Jan‡, James T. Neal‡ *equal contribution ‡co-senior authors
A key challenge of the modern genomics era is developing data-driven representations of gene function. Here, we present the first unbiased morphology-based genome-wide perturbation atlas in human cells, containing three genome-scale genotype-phenotype maps comprising >20,000 single-gene CRISPR-Cas9-based knockout experiments in >30 million cells. Our optical pooled cell profiling approach (PERISCOPE) combines a de-stainable high-dimensional phenotyping panel (based on Cell Painting1,2) with optical sequencing of molecular barcodes and a scalable open-source analysis pipeline to facilitate massively parallel screening of pooled perturbation libraries. This approach provides high-dimensional phenotypic profiles of individual cells, while simultaneously enabling interrogation of subcellular processes. Our atlas reconstructs known pathways and protein-protein interaction networks, identifies culture media-specific responses to gene knockout, and clusters thousands of human genes by phenotypic similarity. Using this atlas, we identify the poorly-characterized disease-associated transmembrane protein TMEM251/LYSET as a Golgi-resident protein essential for mannose-6-phosphate-dependent trafficking of lysosomal enzymes, showing the power of these representations. In sum, our atlas and screening technology represent a rich and accessible resource for connecting genes to cellular functions at scale.
We use conda to manage the computational environment. To install conda see instructions.
After installing conda, execute the following to install and navigate to the environment:
# Install the conda environment
conda env create --force --file environment.yml
# Activate the environment
conda activate periscope_2022Expected install time for the computational environment is less than 10 minutes.
All of our data is publicly available so that our entire workflow can be replicated. However, many of the processing steps require significant computational resources and cannot be replicated on an average personal computer. To address this, we have provided many intermediate files. Input images, intermediate processed images, and the final images used for analysis are available as described in Downloading images. Input profiles, intermediate processed profiles, and the final profiles used for analysis are available as described in Downloading profiles.
Final profile aggregation, hit calling, and all biological discovery described in the paper are performed in Jupyter notebooks provided in this repository. Each folder in this repository contains the input files used, logical intermediates generated, and/or a README file describing how to download any that are too large for the repository.
We have described a minimal example you can use, where possible, to practice using our workflow. However, it is very important to understand that extracting meaningful biology is only possible at scale so any minimal examples are to be used for technical but not biological understanding.
Our workflow should be able to be replicated on any operating system that can run Python and install the dependencies described in our environment.yml and has been tested end-to-end on a Mac operating system and components have been tested on both Windows and Linux operating systems.
Cell Painting images and profiles are available at the Cell Painting Gallery on the Registry of Open Data on AWS (https://registry.opendata.aws/cellpainting-gallery/) under accession number cpg0021-periscope.
Detailed download and file organization information for the Cell Painting Gallery are available at https://github.com/broadinstitute/cellpainting-gallery.
A549 data is labeled as batch 20200805_A549_WG_Screen and may also be referred to as CP186 in some file names.
HeLa data is labeled as batch 20210422_HeLa_WG_Screen and may also be referred to a CP257 in some file names.
DMEM and HPLM conditions were separated by plate.
DMEM plates are 'CP257A','CP257B','CP257D','CP257F', and 'CP257H'.
HPLM_plates are 'CP257J','CP257K','CP257L', and 'CP257N'.
The image metadata will be automatically extracted from the images using the pipelines provided at https://github.com/broadinstitute/pooled-cell-painting-image-processing. An orientation to the images is as follows:
imagesimages contains the raw images as they come off the microscope.
The images can be downloaded using the command
batch=20200805_A549_WG_Screen
plate=CP186A
session=20X_CP_CP186A
aws s3 cp \
--no-sign-request \
--recursive \
s3://cellpainting-gallery/cpg0021-periscope/broad/images/${batch}/images/${plate}/${session} .Within each plate folder, Cell Painting images are organized into a single folder labeled 20XCP${plate}. Each 6-well plate contains images from 1364 (for A549 screen) or 1332 (for HeLa screen) sites for each well. There is 1 .nd2 file per site with five fluorescent images (DAPI, GFP, A594, Cy5, 750) corresponding to the cell painting images. Sites have a 10% overlap.
Within each plate folder, barcoding images are organized into a folder per cycle for each of 12 cycles labeled 10X_c#-SBS-# where # is the cycle number. Each 6-well plate contains images from 320 (for A549 screen) or 316 (for HeLa screen) sites for each well. There is 1 .nd2 file per site with five fluorescent images (DAPI, Cy3, A594, Cy5, Cy7) corresponding to DNA and the four nucleotides read in SBS. Sites have a 10% overlap.
illumillum contains whole plate illumination correction images.
The images can be downloaded using the command
batch=20200805_A549_WG_Screen
plate=CP186A
aws s3 cp \
--no-sign-request \
--recursive \
s3://cellpainting-gallery/cpg0021-periscope/broad/images/${batch}/illum/${plate} .images_alignedimages_aligned contains the whole-plate illumination corrected Barcoding images where channels are aligned within cycles.
Within the images_corrected folder, images are nested by arm (barcoding) and are saved in folders by plate-well.
The images can be downloaded using the command
batch=20200805_A549_WG_Screen
plate=CP186A
aws s3 cp \
--no-sign-request \
--recursive \
s3://cellpainting-gallery/cpg0021-periscope/broad/images/${batch}/images_aligned/barcoding/ . \
--exclude "*" \
--include "${plate}*"images_correctedimages_corrected contains the whole-plate illumination corrected Cell Painting images and the aligned Barcoding images that have undergone channel compensation and background removal.
Within the images_corrected folder, images are nested by arm (barcoding vs. painting) and are saved in folders by plate-well-site.
The images can be downloaded using the command
batch=20200805_A549_WG_Screen
plate=CP186A
aws s3 cp \
--no-sign-request \
--recursive \
s3://cellpainting-gallery/cpg0021-periscope/broad/images/${batch}/images_corrected/ . \
--exclude "*" \
--include "*${plate}*"images_corrected_croppedimages_corrected_cropped contains stitched and cropped pseudo-site images where the pseudo-site corresponds between Cell Painting and Barcoding images.
These are the images used in the final analysis pipeline.
Within the images_corrected folder, images are saved in folders by plate_well.
The images can be downloaded using the command
batch=20200805_A549_WG_Screen
plate=CP186A
aws s3 cp \
--no-sign-request \
--recursive \
s3://cellpainting-gallery/cpg0021-periscope/broad/images/${batch}/images_corrected_cropped/ . \
--exclude "*" \
--include "${plate}*"Image-based profiles were generated using the Pooled Cell Painting Profiling Template and Profiling Recipe. Configuration for A549 profile generation is found in this repository. Configuration for HeLa profile generation is found in this repository.
Find listed below the filenames for all of the profiles generated.
| Dataset | Aggregate_Method | Aggregate_Level | Aggregate_Group | Normalize | Feature_Select | Second_Aggregate_Method | Second_Aggregate_Level | Second_Normalize | Third_Aggregate_Method | Third_Aggregate_Level | File_Name |
|---|---|---|---|---|---|---|---|---|---|---|---|
| A549 | Median | Gene | Plate | N | N | N | N | N | N | N | 20200805_A549_WG_Screen_gene_ALLBATCHES{PLATE}ALLWELLS.csv.gz |
| A549 | Median | Gene | Plate | Standardize | N | N | N | N | N | N | 20200805_A549_WG_Screen_gene_normalized_ALLBATCHES{PLATE}ALLWELLS.csv.gz |
| A549 | Median | Guide | Plate | N | N | N | N | N | N | N | 20200805_A549_WG_Screen_guide_ALLBATCHES{PLATE}ALLWELLS.csv.gz |
| A549 | Median | Guide | Plate | Standardize | N | N | N | N | N | N | 20200805_A549_WG_Screen_guide_normalized_ALLBATCHES{PLATE}ALLWELLS.csv.gz |
| A549 | Median | Guide | Plate | Standardize | N | Median | Y | N | N | N | 20200805_A549_WG_Screen_guide_normalized_median_merged_ALLBATCHES_ALLWELLS.csv.gz |
| HeLa | Median | Guide | Plate | N | N | N | N | N | N | N | 20210422_6W_CP257_guide_ALLBATCHES{PLATE}ALLWELLS.csv |
| HeLa | Median | Gene | Plate | N | N | N | N | N | N | N | 20210422_6W_CP257_gene_ALLBATCHES{PLATE}ALLWELLS.csv |
| HeLa | Median | Gene | Plate | Standardize | N | N | N | N | N | N | 20210422_6W_CP257_gene_normalized_ALLBATCHES{PLATE}ALLWELLS.csv |
| HeLa | Median | Guide | Plate | Standardize | N | N | N | N | N | N | 20210422_6W_CP257_guide_normalized_ALLBATCHES{PLATE}ALLWELLS.csv |
| HeLa | Median | Guide | Plate | Standardize | N | Median | DMEM | N | N | N | 20210422_6W_CP257_guide_normalized_median_merged_ALLBATCHESDMEMALLWELLS.csv |
| HeLa | Median | Guide | Plate | Standardize | N | Median | HPLM | N | N | N | 20210422_6W_CP257_guide_normalized_median_merged_ALLBATCHESHPLMALLWELLS.csv |
Download profiles using the following command
Note that filehead is the filename before the first ___.
dataset=HeLa
filehead=20200805_A549_WG_Screen_gene_ALLBATCHES
aws s3 cp \
--no-sign-request \
s3://cellpainting-gallery/cpg0021-periscope/broad/workspace/profiles/${dataset} . \
--exclude "*" \
--include "${filehead}*"single cell profilesThe datasets also have the additional resource of single cell profiles, available on a per-plate or per-guide basis.
Download per-plate single cell profiles using the following command.
dataset=HeLa
aws s3 cp \
--no-sign-request \
--recursive \
s3://cellpainting-gallery/cpg0021-periscope/broad/workspace/profiles/${dataset}/single_cell/ . \Download per-guide single cell profiles using the following command.
Note that you can enter a specific guide (e.g. AAAAAAAACCCACCTTTCCG) or a gene name (e.g. HSPA8) to download data for all four of its guides.
While single cell by-guide profiles are relatively small, download may be slow as the download command has to list the whole folder.
dataset=HeLa
guide=AAAAAAAACCCACCTTTCCG
aws s3 cp \
--no-sign-request \
--recursive \
s3://cellpainting-gallery/cpg0021-periscope/broad/workspace/profiles/${dataset}/single_cell/ . \
--exclude "*"
--include "*${guide}*"