Uncertainty-aware quantitative analysis of high-throughput live cell migration data

Simo Kitanovski, Shannon Conroy, Justin Sonneck, Lukas Claas, Madeleine Dorsch, Sebastian Urban, Jianxu Chen, Markus Kaiser, Barbara M Grüner and Daniel Hoffmann

PLOS Computational Biology, 2026, vol. 22, issue 7, 1-31

Abstract: Cell migration is a fundamental biological process essential for embryonal development, immune function, and cancer metastasis, with migration velocity representing a key parameter of this behaviour. Today, cell migration velocity can be measured in high-throughput assays that generate complex, hierarchically structured datasets with technical noise, batch effects, and biological variability, introducing significant uncertainty in velocity estimates. Current statistical approaches often fail to rigorously quantify this uncertainty, limiting reproducibility and comparisons across independent experimental datasets. Here, we present cellmig, a specialized computational tool that addresses this challenge. It implements established Bayesian hierarchical modeling within an accessible workflow tailored for high-throughput live cell migration assays, to separate biological signals from technical variation while explicitly quantifying uncertainty in migration velocity. cellmig provides a robust framework for analyzing cell migration assays, including dose-response studies and large-scale screens with multiple biological and technical replicates. By modeling biological variability (e.g., compound-dependent effects) and technical confounders (e.g., batch variability) within a unified Bayesian framework, cellmig estimates condition-specific effects on cell velocity with probabilistic uncertainty intervals, avoiding common pitfalls associated with null-hypothesis testing. Through exhaustive benchmarking against commonly used approaches in the field, we demonstrate that cellmig achieves improved sensitivity in detecting subtle migration effects and enhanced robustness against technical variability. Additionally, its generative models enable simulation of migration velocities under various assumptions, aiding experimental planning. We validated cellmig through a tiered strategy: (1) benchmarking on two independent experimental datasets and (2) deployment on a large-scale high-throughput screen that discovered new chemical biology. Our results demonstrate that cellmig can detect subtle dose-dependent velocity changes, maintain robustness against systematic variability and batch effects, and facilitate reliable integration of multi-experiment datasets. In summary, cellmig enhances reproducibility, reliability, and biological insight in high-throughput migration studies, facilitating quantitative inter-dataset comparisons. cellmig is implemented as an open-source R package and is freely available on Bioconductor (https://bioconductor.org/packages/cellmig).Author summary: Cell migration velocity varies across biological processes–from immune responses to cancer metastasis. Yet accurate measurement remains challenging due to inherent cell-to-cell variability and systematic differences between biological conditions. High-throughput migration assays can track thousands of cells simultaneously under diverse conditions, but they generate complex nested datasets spanning multiple experimental runs with technical and biological replicates. A key challenge is to quantify uncertainty in these multi-experiment datasets, and to reliably detect migration differences. To address this, we developed cellmig: an open-source computational tool that implements probabilistic models. Our approach analyzes cell velocity while rigorously accounting for data hierarchy, batch effects, and noise. It provides robust estimates with uncertainty intervals and even simulates data to optimize experimental design. We demonstrate cellmig’s capabilities using multi-experiment cancer cell migration data. cellmig extracts reliable biological knowledge from cell migration assays and in this way can provide essential information for cancer biology, immunology, and drug development.

Date: 2026
References: Add references at CitEc
Citations:

Downloads: (external link)
https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1014472 (text/html)
https://journals.plos.org/ploscompbiol/article/fil ... 14472&type=printable (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:plo:pcbi00:1014472

DOI: 10.1371/journal.pcbi.1014472

Access Statistics for this article

More articles in PLOS Computational Biology from Public Library of Science
Bibliographic data for series maintained by ploscompbiol ().