Metabolomics

Machine learning models trained on metabolomics data frequently show reduced performance when applied to independent cohorts because of domain shifts introduced by differences in analytical platforms, sample handling, storage conditions, and metabolite quantification procedures. Robust approaches for harmonizing metabolomics datasets across cohorts are therefore needed to improve model transferability and reproducibility. To evaluate the impact of matched samples on domain harmonization and cross-cohort prediction performance in NMR metabolomics data, and to compare multiple harmonization strategies for improving model generalization across independently acquired datasets. A machine-learning framework was developed to classify ICD-10 code I50.9 (unspecified heart failure) using plasma NMR metabolomics data from the Estonian Biobank. Models were trained on a source dataset comprising 12,000 samples measured at Nightingale Health and evaluated on an independently measured destination dataset of 295 samples acquired on a different NMR platform. Harmonization approaches included within-domain normalization, Optimal Transport, Bridge mapping, and Empirical Bayes correction. The effect of matched samples was assessed using up to 108 duplicated biospecimens measured in both datasets. Additional validation was performed using independently generated human gestational diabetes and mouse diabetes metabolomics datasets from Metabolomics Workbench. Domain harmonization generally improved cross-cohort prediction compared with unaligned data, particularly when matched samples were available. Predictive performance increased as the number of matched samples used for calibration increased, although gains varied across methods. Nevertheless, simple within-domain normalization consistently provided strong performance and often matched or exceeded more complex alignment approaches. External validation using fully unmatched human and mouse metabolomics datasets demonstrated that harmonization strategies improved classification performance relative to raw data, supporting the broader applicability of the framework. Matched samples substantially enhance domain harmonization and improve the transferability of metabolomics prediction models across analytical platforms and cohorts. However, robust normalization remains a strong baseline, and the benefits of more sophisticated alignment methods depend on the availability and representativeness of matched reference samples. These findings provide practical guidance for developing reproducible cross-cohort metabolomics machine-learning models while preserving biologically relevant signals.

Retinal organoids have emerged as powerful models of retinal development and disease; however, the extent to which they recapitulate the metabolic landscape of the native retina remains largely unknown. This study aimed to compare the metabolomic profiles of mouse retina and mouse embryonic stem cell-derived retinal organoids and to evaluate metabolic changes during organoid maturation. Untargeted gas chromatography mass spectrometry based metabolomic profiling was performed on intracellular metabolites from mouse retina at postnatal day 3 and mouse retinal organoids (MROs) at differentiation day 28, together with conditioned media collected during organoid maturation. Across mouse retina and retinal organoids, 166 annotated compounds were included in the analysis . Of these, 148 were shared between the two sample types. Using a bidirectional fold-change threshold, differential analysis showed that mouse retinal organoids preserved a substantial metabolic signature shared with mouse retina. However, selective differences were observed in the abundance of glycolysis-associated metabolites and pentose phosphate intermediates including glucose-6-phosphate and ribulose-5-phosphate, as well as tricarboxylic acid-cycle-associated metabolites including citric acid, fumaric acid and oxalic acid. Pathway enrichment analysis identified significant alterations in alanine, aspartate, and glutamate metabolism. Media analysis further revealed stage-dependent changes in extracellular glutamate and glycine abundance during organoid maturation. MRO retains major components of retinal metabolic organization while exhibiting features suggestive of incomplete metabolic maturation compared with native retina. These findings provide valuable insight into MRO metabolic fidelity and guide future optimization strategies.

BACKGROUND:Neuroborreliosis (LNB) poses a significant diagnostic challenge in paediatrics due to its non-specific clinical presentation and the limited sensitivity of standard tests. Metabolomics offers the possibility of direct insight into the biochemical changes occurring in the central nervous system during infection. OBJECTIVE:The aim of the study was to characterise the metabolic profile of paired serum and cerebrospinal fluid (CSF) in children with LNB to identify candidate immunometabolic signatures associated with the infection. METHODS:Samples were analysed using NMR spectroscopy and LC-MS/MS. Data gaps were filled using a Random Forest algorithm, followed by univariate and multivariate statistical tests (PLS-DA) and pathway enrichment analysis based on the KEGG database. RESULTS:Profound dysregulation of purine metabolism was demonstrated, manifested by elevated concentrations of hypoxanthine, xanthine and uric acid, which may reflect a combination of non-specific infection-related responses, oxidative stress, and potentially the pathogen's metabolic demand for host purines. Significant changes were observed in the biosynthesis of aromatic amino acids (particularly L-tyrosine) and in glycerophospholipid metabolism, suggesting damage to the blood-brain barrier. In CSF, alterations in choline and glutamate levels were identified as candidate metabolic signatures potentially associated with neuroinflammation and excitotoxicity. CONCLUSIONS:Metabolomic analysis enables precise mapping of the biochemical response in paediatric LNB. The identified abnormalities in purine metabolism, neuroinflammation and cell membrane integrity represent a promising starting point for future biomarker discovery research.

Post-stroke cognitive impairment (PSCI) is a common long-term complication of acute ischemic stroke often accompanied by metabolic disorder, yet early metabolomic predictors remain poorly characterized. This study aimed to identify acute-phase serum metabolomic signatures associated with PSCI and develop a prediction model for early PSCI risk stratification. In this prospective study, 130 acute ischemic stroke patients were enrolled. Serum samples collected within 24 h of stroke onset were subjected to untargeted metabolomic profiling. Cognitive status was assessed at 3 months after stroke. KEGG pathway and MECNA analyses were performed to assess pathway-level and metabolite-metabolite correlation patterns. Bootstrap-LASSO stability selection and logistic regression were used to develop a prediction model, which was evaluated by stratified 10-fold cross-validation. Calibration curves and decision curve analysis assessed model performance. A total of 51 candidate differential metabolites were identified between PSCI and Non-PSCI patients, involving purine/caffeine-related metabolism, bile acid metabolism, lipid and fatty acid metabolism, and amino acid-related metabolism. MECNA identified 15,120 differential metabolite-metabolite correlation pairs, suggesting distinct acute-phase serum correlation patterns in patients who later developed PSCI. Six metabolites were retained in the final model: 6-Hydroxymellein, 21-Deoxycortisol, Inosine, 2-Hydroxy-3-methylbutyric acid, Isoleucyl-Arginine, and Propylparaben. The metabolite-only model achieved an AUC of 0.774 (95

Pathogenic mitochondrial DNA (mtDNA) mutations contribute to a broad spectrum of both common and rare metabolic diseases. However, clinical presentation is highly variable and only partially explained by the proportion of mutant mtDNA or heteroplasmy. With the relationship between mutation burden and clinical manifestation poorly defined, controlled models are required to uncover underlying mechanisms. Here, we explore the metabolic consequences of increasing heteroplasmy in a well-characterised mouse model harbouring a pathogenic mtDNA deletion. Untargeted urinary metabolomics reveals distinct mutation load-dependent metabolic shifts with some metabolites declining early on, while others exhibit threshold-like increases beyond 60