PNNL

A team of Pacific Northwest National Laboratory (PNNL) researchers have published their paper “Introducing Molecular Hypernetworks for Discovery in Multidimensional Metabolomics Data” in the Journal of Proteome Research.

The paper’s findings represent a significant advancement in the field of metabolomics, which is the study of small molecules produced during cellular metabolism. As the paper says, “Molecular hypernetworks (MHNs) represent a promising tool in the analysis of complex relationships underlying high-dimensional mass spectrometry data.”

MHNs focus on complex multidirectional relationships among molecules, generalizing traditional graph-based methods representing simpler, pairwise relationships. By providing a more comprehensive and efficient way to represent molecular interactions, MHNs have the potential to improve the characterization of biological systems and unknown molecules in untargeted metabolomics studies.

As the authors write, MHNs are an improvement over molecular networks, which are the mathematical graphs that show how different molecules in a sample are related according to their measured properties.

MHNs support “the representation of multiway, multidimensional edges describing connectivity among mass spectrometry features,” the paper says.

The paper describes a method of constructing MHNs directly from existing molecular networks through a process called “clique reconstruction.” This approach allows for the comparison of previously published graph-based molecular networks with their corresponding MHNs, highlighting the advantages of the new method.

The authors write that in future works, they intend to further enhance MHN functionality to develop a “network analysis tool suite” to facilitate the process through which data-informed (bio)chemical interpretations are made.

The PNNL authors include Chief Knowledge Scientist Cliff Joslyn, Data Scientist Andy Lin; Computational Scientist Aivett Bilbao; and Chief Data Scientist and Team Leader for Complex Data Models Emilie Purvine. Coauthor Sean Colby is a senior data scientist at the Open Molecular Software Foundation, Corey Broeckling is a research scientist at Colorado State University, and Madelyn Shapiro is a PhD student at the University of California, Santa Barbara. 

This work was supported by the PNNL Laboratory Directed Research and Development Program and is a contribution of the m/q Initiative. A portion of the research was performed using resources available for research computing at PNNL.