Dr. Alexander Aksenov is an Assistant Professor at the University of Connecticut. His lab focuses on exploring the metabolome of living systems, focusing on the “dark matter” – the unknown unknowns of the system. For more information, please explore the lab’s website: https://aksenovgroup.chemistry.uconn.edu/
This meeting will occur on Tuesday, January 21st, from 12 pm – 1 pm EST.
Title: Illuminating the Dark Matter of Metabolomics Through Molecular Community Networking
Introduction: Molecular networking connects structurally similar metabolites by leveraging MS/MS fragmentation pattern similarities. This approach has enabled a slew of discoveries over the past decade. However, conventional methods rely on arbitrary global spectral similarity thresholds, despite optimal connectivity being molecule class-specific. We present molecular community networking (MCN), an advanced approach that utilizes unpruned full connectivity metabolite networks parsed using network science tools to identify naturally present “molecular communities.” This strategy preserves intra-community connectivity information and optimizes connectivity patterns for each metabolite class, enabling the rescue of lost relationships and the capture of otherwise “hidden” portions of the metabolome.
Methods: Full connectivity metabolite networks were constructed using LC-MS/MS or EI GC- MS data. The Louvain clustering algorithm was employed to identify naturally occurring “molecular communities” within the unpruned networks. Each community was treated as a separate network and pruned using the maximum weight spanning tree algorithm to preserve connectivity while retaining only the most meaningful information. The resulting MCNs represent partitions of the original networks into continua of molecular space, where connections within each molecular family cluster represent the most similar pairs of metabolites across the entire detected metabolome.
Preliminary Data: We validated MCNs using reference spectra and experimental data, demonstrating their ability to assemble molecular space into continua reflecting structural relationships. MCNs rescue lost connectivity between related molecules fractured by conventional networking, for example, link sodiated ion variants to corresponding protonated precursors. We showcase MCN’s utility in discovering new bile acid structures with dipeptide conjugation produced by human microbial cultures, revealing the metabolic capacity of the human microbiota. These molecules were previously undetectable with conventional networks. MCNs exhibit high modularity, suggesting a natural tendency for molecules to group into communities resembling “small-world” structures found in online social networks. This approach empowers molecular discovery in areas such as natural products research, including the reanalysis of existing data to explore previously unconnected molecules.
