CalFluxTools is an R package developed to automate analysis of high-throughput fluorescence recordings of calcium flux across whole microplate assays. The tool is intended for datasets in which oscillatory activity of excitable cells is monitored at plate scale, a format commonly used in neurotoxicity and cardiotoxicity screening. The authors highlight that calcium oscillation profiling can yield more than 60 peak kinetic parameters per well, creating a complex dataset that is difficult to manage manually. CalFluxTools provides an end-to-end pipeline to reduce manual steps and standardize downstream analyses.
CalFluxTools requires three main inputs to run:
These inputs allow the package to associate time-series-derived peak parameters with experimental conditions and to apply user-defined filters before statistical or predictive modeling steps.
The package automates parsing of plate data for 384-well plates and extracts peak kinetic features from the calcium oscillation traces. The source emphasizes that an assay can produce more than 60 distinct peak parameters, and CalFluxTools handles this complexity by computing and organizing these derived features in a standardized way for downstream steps. Specific feature definitions and exact extraction algorithms are not described in detail in the source material; users should consult the package documentation or code repository for implementation specifics.
CalFluxTools generates quality-control metrics and plots to support data review prior to analysis. The QC outputs are intended to help identify plate- or well-level issues that could affect interpretation of derived kinetic parameters. The preprint does not enumerate every QC metric or plot type produced, but the pipeline provides automated visualizations and metrics as part of the standard run.
The package includes built-in statistical analyses such as T-tests and Z-score calculations to compare conditions and identify hits. Additionally, CalFluxTools offers machine-learning–based prediction of compound toxicity values from the extracted peak kinetics. The source notes these capabilities but does not provide detailed information on model types, training procedures, performance metrics, or validation strategies in the provided abstract; such details would be available in the full documentation or repository.
Reported performance in the preprint: CalFluxTools processes a full plate dataset in less than a minute on a MacBook Pro equipped with a 2.6 GHz 6-Core Intel Core i7 processor. The source provides this single benchmark; no additional runtime scaling, memory profiling, or comparisons to alternative tools are reported in the abstract.
The authors provide access to the code repository at https://github.com/ncats/CalFluxTools. The preprint states that the software and supporting data/code links are available with the article. The authors have declared no competing interests. The preprint is released under the terms described on the bioRxiv page; the work is a U.S. Government work and is made available under a CC0 license according to the article metadata.
Funding sources declared in the preprint include the National Institutes of Health, the National Institute on Aging, and the National Center for Advancing Translational Sciences intramural research program (grant identifiers and institutional links are noted in the source). This article is a preprint posted September 21, 2026, on bioRxiv and has not been certified by peer review. For full implementation details, algorithmic parameters, and usage examples, consult the GitHub repository and the supplementary material linked from the preprint page.