An Interpretable Framework to Characterize Compound Treatments on Filamentous Fungi using Cell Painting and Deep Metric Learning
Lejeune, L.; Roussin, M.; Leggio, B.; Vernay, A.
Show abstract
The cell painting microscopy imaging protocol has recently gained traction in the biology community as it allows, through the addition of fluorescent dyes, to acquire images that highlight intra-cellular components that are not visible through traditional whole-cell microscopy. While previous works have successfully applied cell painting to mammalian cells, we devise a staining protocol applicable to a filamentous fungus model. Following a principled visual inspection and annotation protocol of phenotypes by domain-experts, we devise an efficient, robust, and conceptually simple image analysis strategy based on the Deep Cosine Metric Learning paradigm that allows to estimate phenotypical similarities across different imaging modalities. We experimentally demonstrate the benefits of our pipeline in the tasks of estimating dose-response curves over a wide range of subtle phenotypical variations. Last, we showcase how our learned metrics can group image samples according to different modes of action and biological targets in an interpretable manner.
Matching journals
The top 5 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Evaluating discrepancies in dimensionality reduction for time-series single-cell RNA-sequencing data 93%
- A comprehensive comparison on cell type composition inference for spatial transcriptomics data 93%
- SHEST: Single-cell-level artificial intelligence from haematoxylin and eosin morphology for cell type prediction and spatial transcriptomics reconstruction 93%
Similar papers in this journal
- Single-Cell Multi-Modal GAN (scMMGAN) reveals spatial patterns in single-cell data from triple negative breast cancer 95%
- Generating hard-to-obtain information from easy-to-obtain information: applications in drug discovery and clinical inference 94%
- Hierarchical confounder discovery in the experiment-machine learning cycle 92%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.