Monte Carlo randomization tests for large-scale abundance datasets on the GPU

Statistical tests are often performed to discover which experimental variables are reacting to specific treatments. Time-series statistical models usually require the researcher to make assumptions with respect to the distribution of measured responses which may not hold. Randomization tests can be applied to data in order to generate null distributions non-parametrically. However, large numbers of randomizations are required for the precise p-values needed to control false discovery rates. When testing tens of thousands of variables (genes, chemical compounds, or otherwise), significant q-value cutoffs can be extremely small (on the order of 10⁻⁵ to 10⁻⁸). This requires high-precision p-values, which in turn require large numbers of randomizations. The NVIDIA^® Compute Unified Device Architecture^® (CUDA^®) platform for General Programming on the Graphics Processing Unit (GPGPU) was used to implement an application which performs high-precision randomization tests via Monte Carlo sampling for quickly screening custom test statistics for experiments with large numbers of variables, such as microarrays, Next-Generation sequencing read counts, chromatographical signals, or other abundance measurements. The software has been shown to achieve up to more than 12 fold speedup on a Graphics Processing Unit (GPU) when compared to a powerful Central Processing Unit (CPU). The main limitation is concurrent random access of shared memory on the GPU. The software is available from the authors.     

An NMR-based metabolomic assessment of cultured cobia health in response to dietary manipulation

Commercial aquaculture feeds rely heavily on fishmeal and fish oil, which can be expensive and ecologically unsustainable. To evaluate the efficacy of reduced fishmeal diets for outgrowth, a dietary study was conducted on the finfish cobia, Rachycentron canadum. NMR-based metabolomic techniques were used to assess the effect of decreasing dietary fishmeal on the health of the cobia. Filtered serum ¹H NMR spectra analysed by principal components analysis (PCA) showed cobia fed reduced fishmeal diets were metabolically different than cobia on control diets. In particular, tyrosine and betaine increased in cobia fed reduced fishmeal diets while glucose decreased, suggesting that these cobia were not receiving the necessary nutritional components required for energy and growth. The formulated control diet contributed to enriched growth and significantly elevated lactate levels suggesting enhanced gut microflora metabolism in response to dietary components. The results show that NMR-based metabolomic analysis is a useful tool in aquaculture studies.     

Characterization of wines from grape varieties through multivariate statistical analysis of H NMR spectroscopic data

¹H NMR spectroscopic analysis coupled with multivariate statistical data was used to characterize wines vinified from four grape varieties: Muscat Bailey A (Vitis labrusca), Campbell Early (V. labrusca B.), Kyoho (V. labrusca L.) and Meoru (Vitis coignetiae). Pattern recognition methods, such as principal component analysis (PCA) and orthogonal projection to latent structure discriminant analysis (OPLS-DA), showed clear differentiation between wines made from these grape varieties. Metabolites responsible for the differentiation were identified as 2,3-butanediol, glycerol, malate, citrate, tartrate, succinate, lactate, proline, alanine, choline and trigonelline. The PCA score plot of quantitative analysis of targeted profiling data also showed clear separation between the wines. The highest levels of glycerol, 2,3-butanediol, succinate and alcohol were found in Kyoho wines, suggesting higher sugar content in the Kyoho grape berry compared to other grape varieties. Higher contents of citrate and trigonelline in Muscat Bailey A wines, alanine in Campbell Early wines and proline, malate and choline in Meoru wines demonstrated that the metabolites of the wines vary with the grape variety. This study provides insight into the relationship between grape variety and its wine through global and targeted analysis of ¹H NMR spectral data.     

MI-Pack: Increased confidence of metabolite identification in mass spectra by integrating accurate masses and metabolic pathways

Metabolite identification is of central importance to metabolomics as it provides the route to new knowledge. Automated identification of the thousands of peaks detected by high resolution mass spectrometry is currently not possible, largely due to the finite mass accuracy of the spectrometer and the complexity that one peak can be assigned to one or more empirical formula(e) and each formula maps to one or more metabolites. Biological samples are not, however, composed of random metabolite mixtures, but instead comprise of thousands of compounds related through specific chemical transformations. Here we evaluate if prior biological knowledge of these transformations can improve metabolite identification accuracy.Our identification algorithm - which uses metabolite interconnectivity from the KEGG database to putatively identify metabolites by name - is based on mapping an experimentally-derived empirical formula difference for a pair of peaks to a known empirical formula difference between substrate-product pairs derived from KEGG, termed transformation mapping (TM). To maximize identification accuracy, we also developed a novel semi-automated method to calculate a mass error surface associated with experimental peak-pair differences. The TM algorithm with mass error surface has been extensively validated using simulated and experimental datasets by calculating false positive and false negative rates of metabolite identification. Compared to the traditional identification method of database searching accurate masses on a single-peak-by-peak basis, the TM algorithm reduces the false positive rate of identification by > 4-fold, while maintaining a minimal false negative rate. The mass error surface, putative identification of metabolite names, and calculation of false positive and false negative rates collectively advance and improve upon related previous research on this topic [1, 2]. We conclude that inclusion of prior biological knowledge in the form of metabolic pathways provides one route to more accurate metabolite identification.     

Variation in the human lipidome associated with coffee consumption as revealed by quantitative targeted metabolomics

The effect of coffee consumption on human health is still discussed controversially. Here, we report results from a metabolomics study of coffee consumption, where we measured 363 metabolites in blood serum of 284 male participants of the Cooperative Health Research in the Region of Augsburg study population, aged between 55 and 79 years. A statistical analysis of the association of metabolite concentrations and the number of cups of coffee consumed per day showed that coffee intake is positively associated with two classes of sphingomyelins, one containing a hydroxy‐group (SM(OH)) and the other having an additional carboxy‐group (SM(OH,COOH)). In contrast, long‐ and medium‐chain acylcarnitines were found to decrease with increasing coffee consumption. It is noteworthy that the concentration of total cholesterol also rises with an increased coffee intake in this study group. The association observed here between these hydroxylated and carboxylated sphingolipid species and coffee intake may be induced by changes in the cholesterol levels. Alternatively, these molecules may act as scavengers of oxidative species, which decrease with higher coffee intake. In summary, we demonstrate strong positive associations between coffee consumption and two classes of sphingomyelins and a negative association between coffee consumption and long‐ and medium‐chain acylcarnitines.     

Genetic algorithm-based feature selection in high-resolution NMR spectra

High-resolution nuclear magnetic resonance (NMR) spectroscopy has provided a new means for detection and recognition of metabolic changes in biological systems in response to pathophysiological stimuli and to the intake of toxins or nutrition. To identify meaningful patterns from NMR spectra, various statistical pattern recognition methods have been applied to reduce their complexity and uncover implicit metabolic patterns. In this paper, we present a genetic algorithm (GA)-based feature selection method to determine major metabolite features to play a significant role in discrimination of samples among different conditions in high-resolution NMR spectra. In addition, an orthogonal signal filter was employed as a preprocessor of NMR spectra in order to remove any unwanted variation of the data that is unrelated to the discrimination of different conditions. The results of k-nearest neighbors and the partial least squares discriminant analysis of the experimental NMR spectra from human plasma showed the potential advantage of the features obtained from GA-based feature selection combined with an orthogonal signal filter.     

Biomarkers of meat intake and the application of nutrigenomics

Objective dietary intake markers for meat would be useful to assess meat intake in observational studies and as compliance markers in dietary intervention studies. A number of compounds are specific to meat compared with most other dietary items but there is some overlap between protein rich foods. A number of single compounds have been analysed in urine, plasma, serum or hair samples in studies of their relationship to meat or total protein intake. Among potential markers of dietary meat intake are urea, creatine, creatinine, carnitine, carnosine, anserine, ophidine, 1- and 3-methylhistidine, and sulphate or sulphite. Anserine and 1-methylhistidine come close to being meat-specific markers but true quantitative biomarker may not exist. Modern profiling techniques are increasingly used to look for useful biomarkers or for constructing them from latent information in complex profiles. Metabolomics by NMR spectroscopy of urine has also been applied to search for meat intake markers. Studies on single compounds or metabolomics markers are shortly reviewed here.