Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 119 of the 120 references it cites were found in a registry or an index, and 1 are not checked yet.

Unleashing the potential of cell painting assays for compound activities and hazards prediction

Open access
Odje, Floriane, Meijer, David, von Coburg, Elena, van der Hooft, Justin J. J., Dunst, Sebastian, Medema, Marnix H., Volkamer, Andrea · 2024 · Frontiers in Toxicology

Abstract

The cell painting (CP) assay has emerged as a potent imaging-based high-throughput phenotypic profiling (HTPP) tool that provides comprehensive input data for in silico prediction of compound activities and potential hazards in drug discovery and toxicology. CP enables the rapid, multiplexed investigation of various molecular mechanisms for thousands of compounds at the single-cell level. The resulting large volumes of image data provide great opportunities but also pose challenges to image and data analysis routines as well as property prediction models. This review addresses the integration of CP-based phenotypic data together with or in substitute of structural information from compounds into machine (ML) and deep learning (DL) models to predict compound activities for various human-relevant disease endpoints and to identify the underlying modes-of-action (MoA) while avoiding unnecessary animal testing. The successful application of CP in combination with powerful ML/DL models promises further advances in understanding compound responses of cells guiding therapeutic development and risk assessment. Therefore, this review highlights the importance of unlocking the potential of CP assays when combined with molecular fingerprints for compound evaluation and discusses the current challenges that are associated with this approach.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.3389/ftox.2024.1401036

Cites

119 of 120 references verified0 citation errors · 0 not checkable

Statistics

Figures

FIGURE 1 CP in a nutshell. Cells of interest are first cultivated (A) and then seeded to, typically, a 384-well plate (B). Every well is exposed to a single compound or genetic perturbation (E) and incubated for a period ranging from 24 to 48 h, after which every well is uniformly stained with a collection of fluorescent dyes (C). Subsequently, imaging of the cells is carried out (D), and a phenotypic profile is generated for each experimental condition (F). The images of the molecules in section (E.) were created using CineMol (Meijer et al., 2024).
FIGURE 1 CP in a nutshell. Cells of interest are first cultivated (A) and then seeded to, typically, a 384-well plate (B). Every well is exposed to a single compound or genetic perturbation (E) and incubated for a period ranging from 24 to 48 h, after which every well is uniformly stained with a collection of fluorescent dyes (C). Subsequently, imaging of the cells is carried out (D), and a phenotypic profile is generated for each experimental condition (F). The images of the molecules in section (E.) were created using CineMol (Meijer et al., 2024).CC BY · PubMed Central
FIGURE 2 Example of morphological ‘barcode’ representation of CP image data for various concentrations of compound A. Morphological fingerprint includes generic features (G), intensity (I), morphology (M), and texture (T), extracted individually for each channel. Five image channels are depicted: DNA (desoxyribonucleic acid, cyan), RNA (ribonucleic acid) and ER (endoplasmic reticulum, green), AGP (actin, Golgi and plasma membrane, orange), and Mito (mitochondria, red). Features characterize segmented compartments: cells (RNA channel), nuclei (DNA channel), and cytoplasm (excluding nuclei from cells).
FIGURE 2 Example of morphological ‘barcode’ representation of CP image data for various concentrations of compound A. Morphological fingerprint includes generic features (G), intensity (I), morphology (M), and texture (T), extracted individually for each channel. Five image channels are depicted: DNA (desoxyribonucleic acid, cyan), RNA (ribonucleic acid) and ER (endoplasmic reticulum, green), AGP (actin, Golgi and plasma membrane, orange), and Mito (mitochondria, red). Features characterize segmented compartments: cells (RNA channel), nuclei (DNA channel), and cytoplasm (excluding nuclei from cells).CC BY · PubMed Central
FIGURE 3 Compound and cell profile featurization. (A) Compound structures are typically represented as graphs and stored as compressed text, e.g., as SMILES strings. (B) Molecular fingerprint algorithms like a circular Morgan fingerprint can turn the molecular graph structure into an n-bit array, converting the molecular graph into a fixed bit-sized array. Fixed arrays are useful for data analysis and machine learning approaches that require tabular data. (C) The graph structure can also be leveraged directly for graph neural networks, which keep three distinct tables per compound: a table containing atoms and their features, a table containing bonds and their features, and a table that connects the atoms and bonds. (D) Image-based morphological profiles typically consist of multiple grey-scaled images, one image per fluorescent dye used. (E) Segmentation tools can process these image profiles and output a fixed-sized array of features. These morphological fingerprints can equally be leveraged by algorithms that process tabular data. (F) The morphological images can also be used directly in convolutional neural network architectures, which can learn to create a fixed-sized fingerprint, called an embedding.
FIGURE 3 Compound and cell profile featurization. (A) Compound structures are typically represented as graphs and stored as compressed text, e.g., as SMILES strings. (B) Molecular fingerprint algorithms like a circular Morgan fingerprint can turn the molecular graph structure into an n-bit array, converting the molecular graph into a fixed bit-sized array. Fixed arrays are useful for data analysis and machine learning approaches that require tabular data. (C) The graph structure can also be leveraged directly for graph neural networks, which keep three distinct tables per compound: a table containing atoms and their features, a table containing bonds and their features, and a table that connects the atoms and bonds. (D) Image-based morphological profiles typically consist of multiple grey-scaled images, one image per fluorescent dye used. (E) Segmentation tools can process these image profiles and output a fixed-sized array of features. These morphological fingerprints can equally be leveraged by algorithms that process tabular data. (F) The morphological images can also be used directly in convolutional neural network architectures, which can learn to create a fixed-sized fingerprint, called an embedding.CC BY · PubMed Central
FIGURE 4 Using CP data for mechanism of action prediction. (A). Representations for compounds and CP images can be either calculated, for example, with CellProfiler, or learnt with deep learning approaches. (B). Data representations can be concatenated or used independently for downstream tasks. (i) The MoA of new representations can be determined through comparison with similar annotated representations. (ii) Tree models like Random Forest’s can be used to predict MoA for new representations. (iii) Other classical machine learning algorithms like SVMs can also learn to predict MoA for new representations. (iv) Instead of using the representations for downstream tasks, a predictor module can be directly attached to model learning the representations to create an end-to-end predictor.
FIGURE 4 Using CP data for mechanism of action prediction. (A). Representations for compounds and CP images can be either calculated, for example, with CellProfiler, or learnt with deep learning approaches. (B). Data representations can be concatenated or used independently for downstream tasks. (i) The MoA of new representations can be determined through comparison with similar annotated representations. (ii) Tree models like Random Forest’s can be used to predict MoA for new representations. (iii) Other classical machine learning algorithms like SVMs can also learn to predict MoA for new representations. (iv) Instead of using the representations for downstream tasks, a predictor module can be directly attached to model learning the representations to create an end-to-end predictor.CC BY · PubMed Central
FIGURE 5 Flowchart illustrating morphological fingerprint pre-processing with metric examples, including quality control, outlier detection, scaling, and aggregation from single-cell to well to treatment levels, followed by dimensionality reduction for downstream analysis.
FIGURE 5 Flowchart illustrating morphological fingerprint pre-processing with metric examples, including quality control, outlier detection, scaling, and aggregation from single-cell to well to treatment levels, followed by dimensionality reduction for downstream analysis.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 109 references

Research works printed but not linked: 11

  1. Bornholdt, 2022, Deepprofiler
  2. Bray, 2004, Imaging Platform, Broad Institute of MIT and Harvard. Advanced assay development guidelines for image-based high content screening and analysis
  3. 2024, Watershed Segmentation
  4. Carpenter, 2017, Supporting data files, documentation, and updated tips for the Cell Painting protocol, GitHub, 10.1038/nprot.2016.105 · DOI · Paper page
  5. 2023, Globally harmonized system of classification and labelling of chemicals
  6. 2017, Gigadb dataset: id 100351
  7. 2024, Cell painting gallery
  8. Iversen, 2004, HTS assay validation
  9. Josh Moore, 2023, Jumpcellpainting: sample notebook, JUMP-Cell Paint. Consort
  10. Sypetkowski, 2023, Rxrx1: a dataset for evaluating experimental batch correction methods, 4284
  11. Team, 2024, RDKit: open-source cheminformatics, RDKit

Cited by

See all 11

Metadata

Authors
Odje, F., Meijer, D., von Coburg, E., van der Hooft, J. J. J., Dunst, S., Medema, M. H., & Volkamer, A.
Affiliations
  • Saarland University (DE)
  • Wageningen University & Research (NL)
  • German Federal Institute for Risk Assessment (BfR) (DE)
Corresponding author
Volkamer, Andrea
Venue
Frontiers in Toxicology
Published
2024-07-17
Volume
6
Pages
1401036-1401036
ISSN
2673-3080 (ISSN-L)
OpenAlex
W4400745305
Cited by
11
References
109
Citations by year
2026: 5 · 2025: 4 · 2024: 1
FWCI
0.28
Type
journal article
Language
en
Open access
Locations
  • Frontiers in Toxicology (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Socio-Environmental Systems Modeling (submittedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
  • OpenAgrar (submittedVersion, cc-by)
  • Publications of the UdS (Saarland University) (cc-by)
Funders
  • European Research Council
  • Bundesministerium für Bildung und Forschung
Topics
  • Life Sciences › Pharmacology, Toxicology and Pharmaceutics › Drug Discovery › Computational Drug Discovery Methods
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Biophysics › Cell Image Analysis Techniques
  • Health Sciences › Veterinary › Small Animals › Animal testing and alternatives
Keywords
cell painting assay, compound activity prediction, hazard prediction, phenotypic screening, molecular fingerprints, machine learning, deep learning, mechanism of action, image analysis, alternatives to animal testing, drug discovery, toxicology, risk assessment

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Unleashing the potential of cell painting assays for compound activities and hazards prediction. https://app.cochat.ai/papers/10.3389/ftox.2024.1401036
The paper
Odje, F., Meijer, D., von Coburg, E., van der Hooft, J. J. J., Dunst, S., Medema, M. H., & Volkamer, A. (2024). Unleashing the potential of cell painting assays for compound activities and hazards prediction. Frontiers in Toxicology. https://doi.org/10.3389/ftox.2024.1401036

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.