Predicting cell health phenotypes using image-based morphology profiling

Abstract

Genetic and chemical perturbations impact diverse cellular phenotypes, including multiple indicators of cell health. These readouts reveal toxicity and antitumorigenic effects relevant to drug discovery and personalized medicine. We developed two customized microscopy assays, one using four targeted reagents and the other three targeted reagents, to collectively measure 70 specific cell health phenotypes including proliferation, apoptosis, reactive oxygen species (ROS), DNA damage, and cell cycle stage. We then tested an approach to predict multiple cell health phenotypes using Cell Painting, an inexpensive and scalable image-based morphology assay. In matched CRISPR perturbations of three cancer cell lines, we collected both Cell Painting and cell health data. We found that simple machine learning algorithms can predict many cell health readouts directly from Cell Painting images, at less than half the cost. We hypothesized that these models can be applied to accurately predict cell health assay outcomes for any future or existing Cell Painting dataset. For Cell Painting images from a set of 1,500+ compound perturbations across multiple doses, we validated predictions by orthogonal assay readouts. We provide a web app to browse predictions http://broad.io/cell-health-app . Our approach can be used to add cell health annotations to Cell Painting datasets.

ME-VAE: Multi-Encoder Variational AutoEncoder for Controlling Multiple Transformational Features in Single Cell Image Analysis

“…Imaging data has morpho-spatial information that is not captured using a simple mean intensity information of each marker, but successful quantification of these features can lead to improved analysis and understanding. 8 The classical approach for image feature extraction is to manually create a list of desired features and design a metric or algorithm that quantifies those features within the image. This traditional method is biased to known and easily measured features and can miss subtle but important features.…”

BioProfiling.jl: Profiling biological perturbations with high-content imaging in single cells and heterogeneous populations

“…High-Content Screening (HCS) enables profiling cellular phenotypes across hundreds of thousands of conditions by combining automated microscopy with advanced image analysis methods. HCS thus represents a flexible and cost-effective solution for replacing multiple specific assays (Simm et al ., 2018; Way et al ., 2021; Chandrasekaran et al ., 2020), and has been widely adopted in both basic and applied research. Notable achievements range from drug discovery (Simm et al ., 2018; Chandrasekaran et al ., 2020; Scheeder et al ., 2018) to the elucidation of combinatorial drug effects (Caldera et al ., 2019) and to ex-vivo drug-response screening in patients (Snijder et al ., 2017).…”

“…The few actively-maintained tools attempting to fulfill these needs include CellProfiler Analyst and its graphical user interface, designed to handle CellProfiler measurements (McQuin et al ., 2018; Jones et al ., 2008), cellHTS2 in the R programming language, optimized for measurements from plate readers (Boutros et al ., 2006), and more recently cytomapper, an R package for analyzing imaging mass cytometry experiments (Eling et al ., 2021). Of note, efforts are also ongoing to develop general-purpose HCS analysis tools in R and Python, with cytominer and Pycytominer, so far only available as GitHub repositories (Singh, 2021; Way, 2021). Despite the existence of these tools, the field still heavily relies on custom implementation of morphological profile curation for each study to account for different imaging modalities and analytical goals.…”

Brief Guide: Experimental Strategies for High-Quality Hit Selection from Small-Molecule Screening Campaigns

“…20 Profiles based on cell painting can be used to predict and label healthy cells versus compound-mediated cellular toxicity. 21,22 During early drug discovery, we advise running counter, orthogonal, and cellular fitness screens to develop a detailed picture of a compound's effects. This will aid the decision as to whether a compound can be optimized to an effective, specific, and non-cytotoxic lead.…”

“…20 Profiles based on cell painting can be used to predict and label healthy cells versus compound-mediated cellular toxicity. 21,22…”