Disease gene prediction with privileged information and heteroscedastic dropout.

MOTIVATION: Recently, machine learning models have achieved tremendous success in prioritizing candidate genes for genetic diseases. These models are able to accurately quantify the similarity among disease and genes based on the intuition that similar genes are more likely to be associated with similar diseases. However, the genetic features these methods rely on are often hard to collect due to high experimental cost and various other technical limitations. Existing solutions of this problem significantly increase the risk of overfitting and decrease the generalizability of the models. RESULTS: In this work, we propose a graph neural network (GNN) version of the Learning under Privileged Information paradigm to predict new disease gene associations. Unlike previous gene prioritization approaches, our model does not require the genetic features to be the same at training and test stages. If a genetic feature is hard to measure and therefore missing at the test stage, our model could still efficiently incorporate its information during the training process. To implement this, we develop a Heteroscedastic Gaussian Dropout algorithm, where the dropout probability of the GNN model is determined by another GNN model with a mirrored GNN architecture. To evaluate our method, we compared our method with four state-of-the-art methods on the Online Mendelian Inheritance in Man dataset to prioritize candidate disease genes. Extensive evaluations show that our model could improve the prediction accuracy when all the features are available compared to other methods. More importantly, our model could make very accurate predictions when >90% of the features are missing at the test stage. AVAILABILITY AND IMPLEMENTATION: Our method is realized with Python 3.7 and Pytorch 1.5.0 and method and data are freely available at: https://github.com/juanshu30/Disease-Gene-Prioritization-with-Privileged-Information-and-Heteroscedastic-Dropout.

Comprehensive Isotopic Targeted Mass Spectrometry: Reliable Metabolic Flux Analysis with Broad Coverage.

Metabolic flux analysis (MFA) is highly relevant to understanding metabolic mechanisms of various biological processes. While the pace of methodology development in MFA has been rapid, a major challenge the field continues to witness is limited metabolite coverage, often restricted to a small to moderate number of well-known compounds. In addition, isotopic peaks from an enriched metabolite tend to have low abundances, which makes liquid chromatography tandem mass spectrometry (LC-MS/MS) highly useful in MFA due to its high sensitivity and specificity. Previously we have built large-scale LC-MS/MS approaches that can be routinely used for measurement of up to ∼1,900 metabolite/feature levels [Gu et al. Anal. Chem. 2015, 87, 12355-12362. Shi et al. Anal. Chem. 2019, 91, 13737-13745.]. In this study, we aim to expand our previous studies focused on metabolite level measurements to flux analysis and establish a novel comprehensive isotopic targeted mass spectrometry (CIT-MS) method for reliable MFA analysis with broad coverage. As a proof-of-principle, we have applied CIT-MS to compare the steady-state enrichment of metabolites between Myc(oncogene)-On and Myc-Off Tet21N human neuroblastoma cells cultured with U-13C6-glucose medium. CIT-MS is operationalized using multiple reaction monitoring (MRM) mode and is able to perform MFA of 310 identified metabolites (142 reliably detected, 46 kinetically profiled) selected from >35 metabolic pathways of strong biological significance. Further, we developed a novel concept of relative flux, which eliminates the requirement of absolute quantitation in traditional MFA and thus enables comparative MFA under the pseudosteady state. As a result, CIT-MS was shown to possess the advantages of broad coverage, easy implementation, fast throughput, and more importantly, high fidelity and accuracy in MFA. In principle, CIT-MS can be easily adapted to track the flux of other labeled tracers (such as 15N-tracers) in any metabolite detectable by LC-MS/MS and in various biological models (such as mice). Therefore, CIT-MS has great potential to bring new insights to both basic and clinical metabolism research.

Triple Negative Breast Cancer Detection Using LC-MS/MS Lipidomic Profiling.

Breast cancer (BC) is a heterogeneous malignancy that is responsible for a great portion of female cancer cases and cancer-related deaths in the United States. In comparison to other major BC subtypes, triple negative breast cancer (TNBC) presents with a relatively low survival rate and a high rate of metastasis. This has led to a strong, though largely unmet, need for more sensitive and specific methods of early-stage TNBC (ES-TNBC) detection to combat its high-grade pathology and relatively low survival rate. The current study employs a liquid chromatography-tandem mass spectrometry assay capable of targeted, highly specific, and sensitive detection of lipids to propose two diagnostic biomarker panels for TNBC/ES-TNBC. Using this approach, 110 lipids were reliably detected in 166 human plasma samples, 45 controls, and 121 BC (96 non-TNBC and 25 TNBC) subjects. Univariate and multivariate analyses allowed the construction and application of a 19-lipid biomarker panel capable of distinguishing TNBC (and ES-TNBC) from controls, as well as a 5-lipid biomarker panel capable of differentiating TNBC from non-TNBC and ES-TNBC from ES-non-TNBC. Receiver operating characteristic curves with notable classification performances were generated from the biomarker panels according to their orthogonal partial least-squares discrimination analysis models. TNBC was distinguished from controls with an area under the receiving operating characteristic curve (AUROC) = 0.93, sensitivity = 0.96, and specificity = 0.76 and ES-TNBC from controls with an AUROC = 0.96, sensitivity = 0.95, and specificity = 0.89. TNBC was differentiated from non-TNBC with an AUROC = 0.88, sensitivity = 0.88, and specificity = 0.79 and ES-TNBC from ES-non-TNBC with an AUROC = 0.95, sensitivity = 0.95, and specificity = 0.87. A pathway enrichment analysis between TNBC and controls also revealed significant disturbances in choline metabolism, sphingolipid signaling, and glycerophospholipid metabolism. To the best of our knowledge, this is the first study to propose a diagnostic lipid biomarker panel for TNBC detection. All raw mass spectrometry data have been deposited to MassIVE (dataset identifier MSV000085324).

Daily red wine vinegar ingestion for eight weeks improves glucose homeostasis and affects the metabolome but does not reduce adiposity in adults.

Some studies have reported that vinegar ingestion at mealtime attenuates postprandial glycemia in healthy adults and individuals with type 2 diabetes. Emerging data suggest that chronic vinegar ingestion impacts fat metabolism and reduces adiposity, although no study has yet corroborated the events of vinegar supplementation metabolically through a metabolomics approach. To examine the impact of daily vinegar ingestion on glucose homeostasis, adiposity, and the metabolome, an 8-week, randomized controlled trial design was implemented utilizing two parallel treatment arms: daily red wine vinegar ingestion and a control treatment. Participants were 45 healthy adults at increased risk for metabolic complications as determined by high waist circumferences. Measurements and blood samples were collected pre- and post-intervention. Central adiposity and visceral fat were assessed by waist circumference and dual-energy X-ray absorptiometry, respectively. Plasma metabolites were analyzed using gas chromatography-mass spectrometry (MS) and liquid chromatography-MS/MS. Analysis showed significant reductions in fasting glucose (p = 0.003) and insulin (p < 0.001). Insulin resistance was reduced 8.3% in the red wine vinegar group and increased 9.7% in the control group (p < 0.001). No significant between-group differences in body mass index, body weight, waist circumference, or visceral fat were observed. Significant differences were observed in amino valerate and indole-3-acetic acid (p < 0.05), with high magnitudes of fold change (>2) between groups. Metabolic pathway analysis revealed significant alterations in tryptophan metabolism. Although daily red wine vinegar ingestion for 8 weeks induced significant improvements in glucose homeostasis, our results indicate that daily red wine vinegar ingestion for 8 weeks is not associated with reductions in adiposity. This is the first study to investigate the effects of daily red wine vinegar supplementation using a metabolomics approach. Our results provide strong rationales for larger prospective studies to further clarify associations among obesity, chronic diseases, and functional foods such as vinegar using metabolomics.

Statistical analysis and modeling of mass spectrometry-based metabolomics data.

Multivariate statistical techniques are used extensively in metabolomics studies, ranging from biomarker selection to model building and validation. Two model independent variable selection techniques, principal component analysis and two sample t-tests are discussed in this chapter, as well as classification and regression models and model related variable selection techniques, including partial least squares, logistic regression, support vector machine, and random forest. Model evaluation and validation methods, such as leave-one-out cross-validation, Monte Carlo cross-validation, and receiver operating characteristic analysis, are introduced with an emphasis to avoid over-fitting the data. The advantages and the limitations of the statistical techniques are also discussed in this chapter.

Principal component directed partial least squares analysis for combining nuclear magnetic resonance and mass spectrometry data in metabolomics: application to the detection of breast cancer.

Nuclear magnetic resonance (NMR) spectroscopy and mass spectrometry (MS) are the two most commonly used analytical tools in metabolomics, and their complementary nature makes the combination particularly attractive. A combined analytical approach can improve the potential for providing reliable methods to detect metabolic profile alterations in biofluids or tissues caused by disease, toxicity, etc. In this paper, (1)H NMR spectroscopy and direct analysis in real time (DART)-MS were used for the metabolomics analysis of serum samples from breast cancer patients and healthy controls. Principal component analysis (PCA) of the NMR data showed that the first principal component (PC1) scores could be used to separate cancer from normal samples. However, no such obvious clustering could be observed in the PCA score plot of DART-MS data, even though DART-MS can provide a rich and informative metabolic profile. Using a modified multivariate statistical approach, the DART-MS data were then reevaluated by orthogonal signal correction (OSC) pretreated partial least squares (PLS), in which the Y matrix in the regression was set to the PC1 score values from the NMR data analysis. This approach, and a similar one using the first latent variable from PLS-DA of the NMR data resulted in a significant improvement of the separation between the disease samples and normals, and a metabolic profile related to breast cancer could be extracted from DART-MS. The new approach allows the disease classification to be expressed on a continuum as opposed to a binary scale and thus better represents the disease and healthy classifications. An improved metabolic profile obtained by combining MS and NMR by this approach may be useful to achieve more accurate disease detection and gain more insight regarding disease mechanisms and biology.

1H NMR metabolomics study of age profiling in children.

Metabolic profiling of urine provides a fingerprint of personalized endogenous metabolite markers that correlate to a number of factors such as gender, disease, diet, toxicity, medication, and age. It is important to study these factors individually, if possible to unravel their unique contributions. In this study, age-related metabolic changes in children of age 12 years and below were analyzed by (1)H NMR spectroscopy of urine. The effect of age on the urinary metabolite profile was observed as a distinct age-dependent clustering even from the unsupervised principal component analysis. Further analysis, using partial least squares with orthogonal signal correction regression with respect to age, resulted in the identification of an age-related metabolic profile. Metabolites that correlated with age included creatinine, creatine, glycine, betaine/TMAO, citrate, succinate, and acetone. Although creatinine increased with age, all the other metabolites decreased. These results may be potentially useful in assessing the biological age (as opposed to chronological) of young humans as well as in providing a deeper understanding of the confounding factors in the application of metabolomics.

Monitoring diet effects via biofluids and their implications for metabolomics studies.

The effect of diet on metabolites found in rat urine samples has been investigated using nuclear magnetic resonance (NMR) and a new ambient ionization mass spectrometry experiment, extractive electrospray ionization mass spectrometry (EESI-MS). Urine samples from rats with three different dietary regimens were readily distinguished using multivariate statistical analysis on metabolites detected by NMR and MS. To observe the effect of diet on metabolic pathways, metabolites related to specific pathways were also investigated using multivariate statistical analysis. Discrimination is increased by making observations on restricted compound sets. Changes in diet at 24-h intervals led to predictable changes in the spectral data. Principal component analysis was used to separate the rats into groups according to their different dietary regimens using the full NMR, EESI-MS data or restricted sets of peaks in the mass spectra corresponding only to metabolites found in the urea cycle and metabolism of amino groups pathway. By contrast, multivariate analysis of variance from the score plots showed that metabolites of purine metabolism obscure the classification relative to the full metabolite set. These results suggest that it may be possible to reduce the number of statistical variables used by monitoring the biochemical variability of particular pathways. It should also be possible by this procedure to reduce the effect of diet in the biofluid samples for such purposes as disease detection.