Search

High dimensionality omics and hyperspectral imaging datasets present difficult challenges for feature extraction and data mining due to huge numbers of features that cannot be simultaneously examined. The sample numbers and variables of these methods are constantly growing as new technologies are developed, and computational analysis needs to evolve to keep up with growing demand. Current state of the art algorithms can handle some routine datasets but struggle when datasets grow above a certain size. We present a training deep learning via neural networks on non-linear dimensionality reduction, in particular t-distributed stochastic neighbour embedding (t-SNE), to overcome prior limitations of these methods.One Sentence SummaryAnalysis of prohibitively large datasets by combining deep learning via neural networks with non-linear dimensionality reduction.

Diabetes is a global health problem that was estimated to be the 7th leading cause of death worldwide in 2016. Type 2 diabetes mellitus (T2DM) is classically associated with genetic and environmental factors, however recent studies have demonstrated that the gut microbiome, which is altered in T2DM patients, is also likely to play a significant role in disease development. Despite this, the identity of microbiome-derived metabolites that influence T2DM onset and/or progression remain elusive. Here we demonstrate that a serum biomarker for T2DM, previously of unknown structure and origin, is actually two microbiome-derived metabolites, 3-methyl-4-(trimethylammonio)butanoate (3M-4-TMAB) and 4-(trimethylammonio)pentanoate (4-TMAP). These metabolites are produced by the Lachnospiraceae family of bacteria, which are highly prevalent in the gut microbiome of T2DM patients and are associated with high dietary fat intake. Treatment of human liver cells with 3M-4-TMAB and 4-TMAP results in a distinct change in the acylcarnitine profile in these cells and significantly reduced their insulin sensitivity; both indicators of T2DM. These results provide evidence of a mechanistic link between gut microbiome derived metabolites and T2DM.

BackgroundThe gut microbiota is known to influence virtually all facets of human health. Recent work has highlighted a potential role for the gut microbiota in neurological health through the microbiome-gut-brain axis. Microbes can influence the brain both directly and indirectly; through neurotransmitter production, induction of host immunomodulators, or through the release or induction of other microbial or host molecules.MethodsHere we used mass spectrometry imaging (MSI), a label-free imaging tool, to map the molecular changes that occur in the murine gut and brain in germ-free, antibiotic-treated and control mice.ResultsWe determined the spatial distribution and relative quantification of neurotransmitters and their precursors across brain and gut sections in response to the microbiome. Using untargeted MSI of small molecules, we detected a significant change in the levels of four identified metabolites in the brains of germ-free animals compared to controls; vitamin B5, 3-hydroxy-3-methylglutaric acid, 3-methyl-4-(trimethylammonio)butanoate and 4-(trimethylammonio)pentanoate. However, antibiotic treatment induced no significant changes in these metabolites in the brain after one week of treatment.ConclusionsThis work exemplifies the utility of MSI as a tool in determining the spatial distribution and quantification of bacterial and host metabolites in the gut and brain whilst also offering the potential for discovery of novel mediators of microbiome-gut-brain axis communication.

Metabolites, lipids, and other small molecules are key constituents of tissues supporting cellular programs in health and disease. Here, we present METASPACE, a community-populated knowledge base of spatial metabolomes from imaging mass spectrometry data. METASPACE is enabled by a high-performance engine for metabolite annotation in a confidence-controlled way that makes results comparable between experiments and laboratories. By sharing their results publicly, engine users continuously populate a knowledge base of annotated spatial metabolomes in tissues currently including over 3000 datasets from human cancer cohorts, whole-body sections of animal models, and various organs. The spatial metabolomes can be visualized, explored and shared using a web app as well as accessed programmatically for large-scale analysis. By using novel computational methods inspired by natural language processing, we illustrate that METASPACE provides molecular coverage beyond the capacity of any individual laboratory and opens avenues towards comprehensive metabolite atlases on the levels of tissues and organs.