metabodeconplus

A framework for deconvolution, alignment and postprocessing of 1D NMR spectra, resulting in a data matrix of aligned signal integrals that can be used directly for fitting statistical models. The package is described in Schmidt et al. (2026).

Installation

To install the stable version from CRAN, paste the following command in a running R session:

install.packages("metabodeconplus")

To install the development version from GitHub use:

install.packages("pak")
pak::pkg_install("spang-lab/metabodeconplus")

Usage

At Getting Started you can see an example how metabodeconplus can be used to deconvolute an existing data set, followed by alignment of the data and some additional postprocessing steps, resulting in a data matrix of aligned signal integrals.

At Function Reference you get an overview of all functions provided by metabodeconplus.

Documentation

metabodeconplus’s documentation is available at spang-lab.github.io/metabodeconplus. It includes pages about

• Getting Started
• Model Fitting
• Datasets
• Contribution Guidelines
• Function Reference

Related work

metabodeconplus is the latest stage of a package that has been renamed twice:

• MetaboDecon1D (up to v0.2.2, until 2023) was the first implementation, described and evaluated in Haeckl et al. (2021).

• metabodecon (v1.0.0 to v1.7.x, since 2023) is its renamed and reworked successor, which remains available for existing workflows.

• metabodeconplus (v0.20.0 onwards, since 2026) continues metabodecon 1.7 and adds model fitting.

Methods and software used by the package:

• Deconvolution: Koh et al. (2009); optional Rust backend mdrb.

• Alignment: the ‘speaq’ package, described in Beirnaert et al. (2018) and Vu et al. (2011).

• Choice of the number of signals: the Kneedle method of Satopää et al. (2011).

• Classification: random forests (Breiman, 2001) via ‘ranger’ (Wright and Ziegler, 2017) and the lasso via ‘glmnet’ (Friedman et al., 2010).

• 700-bin feature matrix and AKI dataset: Zacharias et al. (2013).

Other tools for 1D NMR processing:

• Deconvolution and profiling: BATMAN (Hao et al., 2012), rDolphin (Cañueto et al., 2018), ASICS (Lefort et al., 2019), BAYESIL (Ravanbakhsh et al., 2015), DEEP Picker1D (Li et al., 2023), decon1d (Hughes et al., 2015), mldecon (Schmid et al., 2023).

• Alignment: icoshift (Savorani et al., 2010), COW (Tomasi et al., 2004).

• Processing workflows: NMRProcFlow (Jacob et al., 2017), SigMa (Khakimov et al., 2020).

• Downstream statistical analysis: MetaboAnalyst (Pang et al., 2024).

Citation

If you use metabodeconplus in your research, please cite:

Schmidt T, Sombke M, Zacharias HU, Oefner PJ, Spang R, Gronwald W, on behalf of the GCKD Investigators. Metabodeconplus—An R Package for Automated Deconvolution and Alignment of 1D NMR Metabolomics Data. Metabolites. 2026;16(9):604. doi:10.3390/metabo16090604