henhaobio
Sign in

bioRxivSystems BiologyPreprint

Identification of potential treatments for COVID-19 through artificial intelligence-enabled phenomic analysis of human cells infected with SARS-CoV-2

Katie HeiseriD, Peter McLeaniD, Chadwick T. Davis, Ben FogelsoniD, Hannah B. Gordon, Pamela Jacobson, Brett L. HurstiD, Ben Miller, Ronald W. Alfa, Berton A. EarnshawiD, Mason L. Victors, Yolanda T. Chong, Imran S. HaqueiD, Adeline S. Low, Christopher C. GibsoniD

PDFDOI
Structured summary available · Jump to summary ↓

Abstract

To identify potential therapeutic stop-gaps for SARS-CoV-2, we evaluated a library of 1,670 approved and reference compounds in an unbiased, cellular image-based screen for their ability to suppress the broad impacts of the SARS-CoV-2 virus on phenomic profiles of human renal cortical epithelial cells using deep learning. In our assay, remdesivir is the only antiviral tested with strong efficacy, neither chloroquine nor hydroxychloroquine have any beneficial effect in this human cell model, and a small number of compounds not currently being pursued clinically for SARS-CoV-2 have efficacy. We observed weak but beneficial class effects of {beta}-blockers, mTOR/PI3K inhibitors and Vitamin D analogues and a mild amplification of the viral phenotype with {beta}-agonists.

Abstract summary

Claude Opus 5.5 · xhigh effort · 18 Oct 2026 · from the abstract · CC BY-NC 4.0

In an image-based screen of 1,670 compounds in SARS-CoV-2-infected human renal epithelial cells, remdesivir was the only antiviral with strong efficacy; chloroquine and hydroxychloroquine showed no benefit.

A stop-gap drug for SARS-CoV-2 had to be found among approved and reference compounds, which calls for a readout that ranks them without assuming a mechanism. Imaging infected human cells and scoring the phenomic profiles with deep learning provides one, and puts chloroquine and hydroxychloroquine to the same test as everything else in the library.

Key results

  1. SuggestedRemdesivir was the one antiviral in the library that strongly reversed the infection-associated cell phenotype, while chloroquine and hydroxychloroquine gave no benefit in this human cell model.

    1,670 approved and reference compounds screened in SARS-CoV-2-infected human renal cortical epithelial cells; the abstract reports the outcome qualitatively, with no dose-response, effect size or significance values. · Abstract

  2. ClaimedA handful of compounds outside the clinical programmes then running for SARS-CoV-2 also scored as effective.

    Stated as a small number of hits, none of which is named in the abstract, and no validation in a second assay or cell type is reported. · Abstract

  3. SuggestedSeveral drug classes moved the phenotype as a group: beta-blockers, mTOR/PI3K inhibitors and vitamin D analogues weakly towards recovery, beta-agonists mildly towards a stronger viral phenotype.

    Reported as weak class effects and a mild amplification respectively; no per-compound values, class sizes or statistics are given. · Abstract

Methods, data and reproducibility

Unbiased cellular imaging of human renal cortical epithelial cells infected with SARS-CoV-2: deep learning turns the images into phenomic profiles, and each compound is scored by how far it suppresses the broad changes the virus produces, rather than against a predefined mechanism.

  • Library of approved and reference compounds screened against SARS-CoV-2 infection phenotypes in renal cortical epithelial cells · human · n = 1,670 compounds
Reproducibility gaps
No compound list, image data, model weights or code are pointed to, and the scoring procedure and thresholds are not described.
Limitations (2)
  • One human cell type and one imaging readout stand in for infection in patients, and no second cell model or orthogonal viral assay is reported. · noted in this summary
  • Effects are described in words with no concentrations, effect sizes or statistics, and the promising non-clinical compounds are not named, so nothing can be followed up from the abstract. · noted in this summary

Read this if you triage repurposing candidates by morphological profiling, or want a human-cell comparison of remdesivir against the chloroquines.

Citations
68

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISBiering '20Ko '20Nguyenla '20Yao '23Liu '20Bakowski '21
68% semantic matchMicrobiology
S. B. Biering, E. Van Dis, E. Wehri+20 more·bioRxiv·Dec 30, 2020