RESUMEN
Non-targeted metabolomics via high-resolution mass spectrometry methods, such as direct infusion Fourier transform-ion cyclotron resonance mass spectrometry (DI-FT-ICR MS), produces data sets with thousands of features. By contrast, the number of samples is in general substantially lower. This disparity presents challenges when analyzing non-targeted metabolomics data sets and often requires custom methods to uncover information not always accessible via classical statistical techniques. In this work, we present a pipeline that combines a convolutional neural network with traditional statistical approaches and an adaptation of a genetic algorithm. The developed method was applied to a lifestyle intervention cohort data set, where subjects at risk of type 2 diabetes underwent an oral glucose tolerance test. Feature selection is the final result of the pipeline, achieved through classification of the data set via a neural network, with a precision-recall score of over 0.9 on the test set. The features most relevant for the described classification were then chosen via a genetic algorithm. The output of the developed pipeline encompasses approximately 200 features with high predictive scores, providing a fingerprint of the metabolic changes in the prediabetic class on the data set. Our framework presents a new approach which allows to apply complex modeling based on convolutional neural networks for the analysis of high-resolution mass spectrometric data.
Asunto(s)
Diabetes Mellitus Tipo 2 , Humanos , Espectrometría de Masas/métodos , Metabolómica/métodos , Redes Neurales de la ComputaciónRESUMEN
BACKGROUND: To study the spatial heterogeneity of liver fat fraction changes during a long-term lifestyle intervention study using magnetic resonance imaging (MRI). METHODS: Thirty-two subjects underwent two MRI-scans in a span of one year. A chemical shift encoding-based water-fat separation method was applied to measure liver proton density fat fraction (PDFF) maps. The PDFF changes in the two liver lobes and the Couinaud segments were compared with the mean liver PDFF change. RESULTS: The slope of the relationship between mean liver PDFF changes and PDFF liver lobe changes was higher in the right compared to the left lobe (slopemean PDFF whole liver ~ mean PDFF right lobe =1.08, slopemean PDFF whole liver ~ mean PDFF left lobe =0.93, P<0.001). The highest slope of agreement between PDFF changes in each specific liver segment and mean liver PDFF changes was observed in segment VII (slope =1.12). The lowest slope of agreement between PDFF changes in each specific liver segment and mean liver PDFF changes was observed in segment I (slope =0.77). CONCLUSIONS: Larger PDFF changes in the right liver lobe were observed compared to PDFF changes in the left liver lobe (LLL) in subjects with both increasing and decreasing mean liver PDFF after one year. The results are in line with the existing literature reporting a heterogeneous spatial distribution of liver fat and highlight the need to spatially resolve liver fat fraction changes in longitudinal studies.
RESUMEN
Variants in FTO have the strongest association with obesity; however, it is still unclear how those noncoding variants mechanistically affect whole-body physiology. We engineered a deletion of the rs1421085 conserved cis-regulatory module (CRM) in mice and confirmed in vivo that the CRM modulates Irx3 and Irx5 gene expression and mitochondrial function in adipocytes. The CRM affects molecular and cellular phenotypes in an adipose depot-dependent manner and affects organismal phenotypes that are relevant for obesity, including decreased high-fat diet-induced weight gain, decreased whole-body fat mass, and decreased skin fat thickness. Last, we connected the CRM to a genetically determined effect on steroid patterns in males that was dependent on nutritional challenge and conserved across mice and humans. Together, our data establish cross-species conservation of the rs1421085 regulatory circuitry at the molecular, cellular, metabolic, and organismal level, revealing previously unknown contextual dependence of the variant's action.