All data are publicly available.
Data | Level | Location | Notes |
---|---|---|---|
Images | 1 | Image Data Resource (IDR) | Accession idr0013(screenA) |
This repository is structured as follows:
Order | Module | Description |
---|---|---|
0.locate_data | Locate mitosis movies | Find locations (plate, well, frame) for training and control movies |
1.idr_streams | Extract features from mitosis movies | Use idrstream to extract features from training and control movies |
2.format_training_data | Format training data | Compile metadata, phenotypic class, and feature data for Mitocheck-labeled movies |
3.normalize_data | Normalize data | Use UMAP to suggest batch effects are not dominant signal and normalize with data using negative controls as normalization population |
4.analyze_data | Analyze data | Analyze normalized data |
Other necessary folders/files:
Folder/File | Description |
---|---|
mitocheck_metadata | IDR curated metadata, trainingset file and features dataset necessary for locating Mitocheck-labeled training data |
utils | Python files with functions used throughout repository |
mitocheck_data_env.yml | Environment file with packages necessary to process mitocheck data |
As part of the Phenotypic profiling of the human genome by time-lapse microscopy reveals cell division genes, Mitocheck created a training dataset with cell phenotypic classes and their locations. This dataset was provided by J.K. Hériché and is located in mitocheck_metadata. This dataset contains the following files:
- trainingset.dat : Mitocheck-assigned object IDs and phenotypic class for cells from a specified frame, well, and plate.
- features/ : Mitocheck-assigned object IDs and bounding boxes for cells from a specified frame, well, and plate.
We use trainingset.dat
to locate the frame, well, and plate of labeled cells in 0.locate_data.
After extracting the features from these labeled frames with idrstream
, we associate the bounding boxes of cells from features/
with their idrstream
-derived coordinates to assign cells their phenotypic class (as labeled by Mitocheck).
We extract single-cell features from positive and negative controls, which are useful for normalizing all Mitocheck data and suggesting that batch effects are not a dominant signal.
We use IDR-curated mitocheck metadata to locate the well and plate of each control movie.
Because idrstream
can only extract features from a single frame, we choose a random frame from the middle 33% of the movie.
Mitocheck mitosis movies are about 93 frames long, so a random frame between frames 31 and 62 are chosen to extract features from.
Because we cannot exactly align the movies in time, we opt to randomly sample from the middle of the movies.
We extract all single-cell features with and without illumination correction.
We refer to these dataset types in the code as dataset_type
, where ic
corresponds to the dataset with illumination correction and no_ic
corresponds to the dataset without illumination correction.
Perform the following steps to set up the mitocheck_data
environment necessary for processing data in this repository.
# Run this command to create the conda environment for mitocheck data processing
conda env create -f mitocheck_data_env.yml
# Run this command to activate the conda environment for mitocheck data processing
conda activate mitocheck_data
In this module, we collect and package data created by this project.
This module leverages system-available Python, Poetry, and Poe the Poet (among other dependencies found in the pyproject.toml
file) to complete tasks.
This module also leverages Docker to reproducibly leverage additional tooling outside of Python dependencies.
We recommend installing Docker (suggested through Docker Desktop), Python (suggested through pyenv) and Poetry (suggested through pip install poetry
), then using the following to run the processes related to this step.
# after installing poetry, create the environment
poetry install
# run the poe the poet task related to this step
# (triggers multiple Python modules)
poetry run poe package_data