Phenotypic Profiling of High Throughput Imaging Screens with Generic Deep Convolutional Features

Abstract

While deep learning has seen many recent applications to drug discovery, most have focused on predicting activity or toxicity directly from chemical structure. Phenotypic changes exhibited in cellular images are also indications of the mechanism of action (MoA) of chemical compounds. In this paper, we show how pre-trained convolutional image features can be used to assist scientists in discovering interesting chemical clusters for further investigation. Our method reduces the dimensionality of raw fluorescent stained images from a high throughput imaging (HTI) screen, producing an embedding space that groups together images with similar cellular phenotypes. Running standard unsupervised clustering on this embedding space yields a set of distinct phenotypic clusters. This allows scientists to further select and focus on interesting clusters for downstream analyses. We validate the consistency of our embedding space qualitatively with t-sne visualizations, and quantitatively by measuring embedding variance among images that are known to be similar. Results suggested the usefulness of our proposed workflow using deep learning and clustering and it can lead to robust HTI screening and compound triage.

Multiplexed high-throughput immune cell imaging reveals molecular health-associated phenotypes

“…Whilst this is not the first work which deploys CNN-based cell classification (Moen et al 2019; Kraus et al 2017; Kraus, Ba, and Frey 2016; Pärnamaa and Parts 2017; Dürr and Sick 2016; Sommer et al 2017; Kandaswamy et al 2016; Godinez et al 2017; Hussain et al 2019) and feature extraction (Pärnamaa and Parts 2017; Kraus et al 2017; Jackson et al 2019; Godinez et al 2017), to our knowledge, this is the first work where deep learning is applied in high-throughput screening and phenotypic analyses of primary human PBMCs. By training the CNN on curated cells from across independent experiments, multiple donors, and conventional and multiplexed staining panels, we could prevent overfitting on phenotypes of single donors and technical bias stemming from experimental conditions.…”

Spectrum of deep learning algorithms in drug discovery

“…They also claimed that ATNC could effectively produce hit compounds (Putin, Asadulaev, Vanhaelen, Ivanenkov, & Aladinskaya, 2018). In another aspect, clustering of cellular images using the CNN approach in HTS workflow has made it easier to find hit compounds in screening activities (Jackson et al, 2019). A strategy involving SVM linear combinations of differentially weighted models with the ability to differentiate between desired activity profiles and the undesired ones has also been introduced by Heikamp and Bajorath, while standard SVM models were unable to differentiate between the compounds with distinct activity profiles and compounds with overlapping activity profiles (Heikamp & Bajorath, 2013).…”

Cell Morphology-Guided De Novo Hit Design by Conditioning Generative Adversarial Networks on Phenotypic Image Features

“…Cell Painting uses different dyes to simultaneously stain several organelles and cellular components, thus capturing information of the complete cellular state 18 . In the context of profiling, images are usually processed by computational pipelines to extract feature representations (morphological profiles) [19][20][21][22][23] , which serve as phenotype descriptors in further tasks. This technique has been used to cluster small molecules by similar phenotypic effect 18 , for drug repurposing 24 , to map cellular morphology to gene function 25 , and to predict results from biochemical assays 26,27 among other applications 28 .…”