Self-supervision advances morphological profiling by unlocking powerful image representations

Abstract Cell Painting is an image-based assay that offers valuable insights into drug mechanisms of action and off-target effects. However, traditional feature extraction tools such as CellProfiler are computationally intensive and require frequent parameter adjustments. Inspired by recent advances...

Full description

Saved in:
Bibliographic Details
Main Authors: Vladislav Kim, Nikolaos Adaloglou, Marc Osterland, Flavio M. Morelli, Marah Halawa, Tim König, David Gnutt, Paula A. Marin Zapata
Format: Article
Language:English
Published: Nature Portfolio 2025-02-01
Series:Scientific Reports
Online Access:https://doi.org/10.1038/s41598-025-88825-4
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850198058944430080
author Vladislav Kim
Nikolaos Adaloglou
Marc Osterland
Flavio M. Morelli
Marah Halawa
Tim König
David Gnutt
Paula A. Marin Zapata
author_facet Vladislav Kim
Nikolaos Adaloglou
Marc Osterland
Flavio M. Morelli
Marah Halawa
Tim König
David Gnutt
Paula A. Marin Zapata
author_sort Vladislav Kim
collection DOAJ
description Abstract Cell Painting is an image-based assay that offers valuable insights into drug mechanisms of action and off-target effects. However, traditional feature extraction tools such as CellProfiler are computationally intensive and require frequent parameter adjustments. Inspired by recent advances in AI, we trained self-supervised learning (SSL) models DINO, MAE, and SimCLR on a subset of the JUMP Cell Painting dataset to obtain powerful representations for Cell Painting images. We assessed these SSL features for reproducibility, biological relevance, predictive power, and transferability to novel tasks and datasets. Our best model (DINO) surpassed CellProfiler in drug target and gene family classification, significantly reducing computational time and costs. DINO showed remarkable generalizability without fine-tuning, outperforming CellProfiler on an unseen dataset of genetic perturbations. In bioactivity prediction, DINO achieved comparable performance to models trained directly on Cell Painting images, with only a small gap between supervised and self-supervised approaches. Our study demonstrates the effectiveness of SSL methods for morphological profiling, suggesting promising research directions for improving the analysis of related image modalities.
format Article
id doaj-art-3bc41c92a76c4a72b1013e9f4b34884a
institution OA Journals
issn 2045-2322
language English
publishDate 2025-02-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-3bc41c92a76c4a72b1013e9f4b34884a2025-08-20T02:12:58ZengNature PortfolioScientific Reports2045-23222025-02-0115111510.1038/s41598-025-88825-4Self-supervision advances morphological profiling by unlocking powerful image representationsVladislav Kim0Nikolaos Adaloglou1Marc Osterland2Flavio M. Morelli3Marah Halawa4Tim König5David Gnutt6Paula A. Marin Zapata7Machine Learning Research, Bayer AGMachine Learning Research, Bayer AGMachine Learning Research, Bayer AGMachine Learning Research, Bayer AGMachine Learning Research, Bayer AGImage-Based Screening Systems, Bayer AGImage-Based Screening Systems, Bayer AGMachine Learning Research, Bayer AGAbstract Cell Painting is an image-based assay that offers valuable insights into drug mechanisms of action and off-target effects. However, traditional feature extraction tools such as CellProfiler are computationally intensive and require frequent parameter adjustments. Inspired by recent advances in AI, we trained self-supervised learning (SSL) models DINO, MAE, and SimCLR on a subset of the JUMP Cell Painting dataset to obtain powerful representations for Cell Painting images. We assessed these SSL features for reproducibility, biological relevance, predictive power, and transferability to novel tasks and datasets. Our best model (DINO) surpassed CellProfiler in drug target and gene family classification, significantly reducing computational time and costs. DINO showed remarkable generalizability without fine-tuning, outperforming CellProfiler on an unseen dataset of genetic perturbations. In bioactivity prediction, DINO achieved comparable performance to models trained directly on Cell Painting images, with only a small gap between supervised and self-supervised approaches. Our study demonstrates the effectiveness of SSL methods for morphological profiling, suggesting promising research directions for improving the analysis of related image modalities.https://doi.org/10.1038/s41598-025-88825-4
spellingShingle Vladislav Kim
Nikolaos Adaloglou
Marc Osterland
Flavio M. Morelli
Marah Halawa
Tim König
David Gnutt
Paula A. Marin Zapata
Self-supervision advances morphological profiling by unlocking powerful image representations
Scientific Reports
title Self-supervision advances morphological profiling by unlocking powerful image representations
title_full Self-supervision advances morphological profiling by unlocking powerful image representations
title_fullStr Self-supervision advances morphological profiling by unlocking powerful image representations
title_full_unstemmed Self-supervision advances morphological profiling by unlocking powerful image representations
title_short Self-supervision advances morphological profiling by unlocking powerful image representations
title_sort self supervision advances morphological profiling by unlocking powerful image representations
url https://doi.org/10.1038/s41598-025-88825-4
work_keys_str_mv AT vladislavkim selfsupervisionadvancesmorphologicalprofilingbyunlockingpowerfulimagerepresentations
AT nikolaosadaloglou selfsupervisionadvancesmorphologicalprofilingbyunlockingpowerfulimagerepresentations
AT marcosterland selfsupervisionadvancesmorphologicalprofilingbyunlockingpowerfulimagerepresentations
AT flaviommorelli selfsupervisionadvancesmorphologicalprofilingbyunlockingpowerfulimagerepresentations
AT marahhalawa selfsupervisionadvancesmorphologicalprofilingbyunlockingpowerfulimagerepresentations
AT timkonig selfsupervisionadvancesmorphologicalprofilingbyunlockingpowerfulimagerepresentations
AT davidgnutt selfsupervisionadvancesmorphologicalprofilingbyunlockingpowerfulimagerepresentations
AT paulaamarinzapata selfsupervisionadvancesmorphologicalprofilingbyunlockingpowerfulimagerepresentations