Cell Health - Cell Painting Single Cell Profiles
Datasets usually provide raw data for analysis. This raw data often comes in spreadsheet form, but can be any collection of data, on which analysis can be performed.
Single Cell Databases of Cell Painting Profiles for the Cell Health Project. These data are used to aggregate profiles in a CRISPR knockout experiment. The data are used to predict cell health assays.
We collected Cell Painting measurements on a CRISPR experiment. The experiment targeted 59 genes, which included 119 unique guides (~2 per gene), across 3 cell lines. The cell lines included A549, ES2, and HCC44.
About 40% of all CRISPR guides were reproducible. This is ok since we are not actually interested in the CRISPR treatment specifically, but instead, just its corresponding readout in each cell health assay.
We performed the following approach:
Split data into 85% training and 15% test sets.
Normalized data by plate (z-score).
Selected optimal hyperparamters using 5-fold cross-validation
Trained elastic net regression models to predict each of the 70 cell health assay readouts, independently.
Trained using shuffled data as well.
Report performance on training and test sets.
We also trained logistic regression classifiers using the same approach above
See https://github.com/broadinstitute/cell-health for more details.
Advancing algorithms for image-based profiling
National Institute of General Medical SciencesFind out more...
Select an IC:
- CA - National Cancer Institute (NCI)