Skip to main content

nmRanalysis: New Software Leverages Machine Learning to Analyze Cellular Metabolites

A user-friendly web application leverages the power of machine learning to improve efficiency and reproducibility of identifying and measuring the concentrations of metabolites in nuclear magnetic resonance spectra. 

circular data streams

A team of researchers from the Environmental Molecular Sciences Laboratory and Pacific Northwest National Laboratory developed nmRanalysis as a user-friendly, open-source web application that streamlines the chemical compound profiling process. (Graphic by koto_feja | iStock)

The Science  

Nuclear magnetic resonance (NMR) spectroscopy is a powerful method for analyzing chemical mixtures. However, identifying and quantifying (i.e. “profiling”) their composition, especially for complex mixtures of biological samples, remains a major challenge. While there are many existing software tools for profiling NMR spectra, they all suffer from a wide variety of obstacles. For example, some lack the ability to handle automation and others require a great deal of technical expertise, or don’t have an easy-to-use graphical user interface. Still, others are limited in the variety of NMR techniques they can be applied to; or they may be unable to handle large volumes of data. To address all of these challenges, nmRanalysis was developed as a user-friendly, open-source web application that streamlines the profiling process. Built with R Shiny, a web application framework for the R programming language, it integrates open-source tools and databases while introducing a fast, machine learning-based recommender system for compound identification. Designed for broad applicability, nmRanalysis improves efficiency and accessibility in NMR data analysis without requiring programming skills. 

Flowchart of nmRanalysis process, featuring steps like uploading data, visualizing, and interacting with predictive models, plus options to download and modify fitting parameters, connected by directional arrows.
Overview of the nmRanalysis workflow: nuclear magnetic resonance data are input (A), processed (B), and modeled to predict metabolite presence/absence (C); then settings are adjusted for quantification (D), metabolites are quantified (E), and results are visualized (F). (Image courtesy of the nmRanalysis development team | Pacific Northwest National Laboratory)

The Impact 

NMR spectroscopy is often used to determine which small molecules (i.e., metabolites) are present and how much of them are found in biological samples. This information, aka the molecular profile, is valuable across many fields—from medicine and pharmaceuticals to biofuels and environmental science. Traditionally, molecular profiling can be slow and requires extensive manual curation due to complex, noisy signals. Now, a team of researchers from Pacific Northwest National Laboratory (PNNL) and the Environmental Molecular Sciences Laboratory (EMSL), a DOE Office of Science user facility located at PNNL, accelerated molecular profiling of biological samples by using data from previous experiments with biological samples to teach machine learning models to accurately profile NMR spectra. These models are packaged within a software application to make the technology more accessible, removing the most significant bottleneck in metabolic profiling and enabling more rapid scientific discovery.  

Summary 

Though data acquisition and initial signal preprocessing of NMR spectra have achieved high degrees of automation, downstream processing—specifically the profiling of spectra—has constricted the overall NMR analysis workflow. Several efforts have been made to mitigate this bottleneck, but these solutions often trade an increase in automation for limitations elsewhere. In this work, a team of PNNL and EMSL researchers introduce nmRanalysis, a user-friendly web application that integrates the strengths of existing profiling tools for a more automated profiling workflow. nmRanalysis additionally incorporates novel features, including a machine learning-driven recommender system for metabolite identification which further increases the utility of nmRanalysis over the individual tools that it incorporates. 

Contacts  

Javier E. Flores 

PNNL 

javier.flores@pnnl.gov  

William Kew 

EMSL | PNNL 

William.kew@pnnl.gov  

Funding 

This research was performed on an EMSL Intramural Science and Technology project award from the Environmental Molecular Sciences Laboratory, a DOE Office of Science user facility at PNNL sponsored by the Biological and Environmental Research program. 

Publication 

J.E. Flores, et al. “nmRanalysis: An open-source web application for semi-automated NMR data analysis.” Analytical Chemistry 97(13), 7037–7046 (2025). [DOI: 10.1021/acs.analchem.4c05104