A Partial Least Squares based algorithm for parsimonious variable selection

Prediction of cyanidin 3-rutinoside content in Michelia crassipes based on near-infrared spectroscopic techniques

Currently the determination of cyanidin 3-rutinoside content in plant petals usually requires chemical assays or high performance liquid chromatography (HPLC), which are time-consuming and laborious. In this study, we aimed to develop a low-cost, high-throughput method to predict cyanidin 3-rutinoside content, and developed a cyanidin 3-rutinoside prediction model using near-infrared (NIR) spectroscopy combined with partial least squares regression (PLSR). We collected spectral data from Michelia crassipes (Magnoliaceae) tepals and used five different preprocessing methods and four variable selection algorithms to calibrate the PLSR model to determine the best prediction model. The results showed that (1) the PLSR model built by combining the blockScale (BS) preprocessing method and the Significance multivariate correlation (sMC) algorithm performed the best; (2) The model has a reliable prediction ability, with a coefficient of determination (R2) of 0.72, a root mean square error (RMSE) of 1.04%, and a residual prediction deviation (RPD) of 2.06. The model can be effectively used to predict the cyanidin 3-rutinoside content of the perianth slices of M. crassipes, providing an efficient method for the rapid determination of cyanidin 3-rutinoside content.

Biogeographical ancestry, variable selection, and PLS-DA method: a new panel to assess ancestry in forensic samples via MPS technology

As evidenced by the large number of articles recently published in the literature, forensic scientists are making great efforts to infer externally visible features and biogeographical ancestry (BGA) from DNA analysis. Just as phenotypic, ancestry information obtained from DNA can provide investigative leads to identify the victims (missing/unidentified persons, crime/armed conflict/mass disaster victims) or trace their perpetrators when no matches were found with the reference profile or in the database. Recently, the advent of Massively Parallel Sequencing technologies associated with the possibility of harnessing high-throughput genetic data allowed us to investigate the associations between phenotypic and genomic variations in worldwide human populations and develop new BGA forensic tools capable of simultaneously analyzing up to millions of markers if for example the ancient DNA approach of hybridization capture was adopted to target SNPs of interest. In the present study, a selection of more than 3000 SNPs was performed to create a new BGA panel and the accuracy of the new panel to infer ancestry from unknown samples was evaluated by the PLS-DA method. Subsequently, the panel created was assessed using three variable selection techniques (Backward variable elimination, Genetic Algorithm and Regularized elimination procedure), and the best SNPs in terms of inferring bio-geographical ancestry at inter- and intra-continental level were selected to obtain panels to predict BGA with a reduced number of selected markers to be applied in routine forensic cases where PCR amplification is the best choice to target SNPs.

Prediction of high fecal propionate-to-butyrate ratios using 16S rRNA-based detection of bacterial groups with liquid array diagnostics

Butyrate and propionate represent two of three main short-chain fatty acids produced by the intestinal microbiota. In healthy populations, their levels are reportedly equimolar, whereas a deviation in their ratio has been observed in various diseased cohorts. Monitoring such a ratio represents a valuable metric; however, it remains a challenge to adopt short-chain fatty acid detection techniques in clinical settings because of the volatile nature of these acids. Here we aimed to estimate short-chain fatty acid information indirectly through a novel, simple quantitative PCR-compatible assay (liquid array diagnostics) targeting a limited number of microbiome 16S markers. Utilizing 15 liquid array diagnostics probes to target microbiome markers selected by a model that combines partial least squares and linear discriminant analysis, the classes (normal vs high propionate-to-butyrate ratio) separated at a threshold of 2.6 with a prediction accuracy of 96%.

Quantitative Comparison of Statistical Methods for Analyzing Human Metabolomics Data

Emerging technologies now allow for mass spectrometry-based profiling of thousands of small molecule metabolites ('metabolomics') in an increasing number of biosamples. While offering great promise for insight into the pathogenesis of human disease, standard approaches have not yet been established for statistically analyzing increasingly complex, high-dimensional human metabolomics data in relation to clinical phenotypes, including disease outcomes. To determine optimal approaches for analysis, we formally compare traditional and newer statistical learning methods across a range of metabolomics dataset types. In simulated and experimental metabolomics data derived from large population-based human cohorts, we observe that with an increasing number of study subjects, univariate compared to multivariate methods result in an apparently higher false discovery rate as represented by substantial correlation between metabolites directly associated with the outcome and metabolites not associated with the outcome. Although the higher frequency of such associations would not be considered false in the strict statistical sense, it may be considered biologically less informative. In scenarios wherein the number of assayed metabolites increases, as in measures of nontargeted versus targeted metabolomics, multivariate methods performed especially favorably across a range of statistical operating characteristics. In nontargeted metabolomics datasets that included thousands of metabolite measures, sparse multivariate models demonstrated greater selectivity and lower potential for spurious relationships. When the number of metabolites was similar to or exceeded the number of study subjects, as is common with nontargeted metabolomics analysis of relatively small cohorts, sparse multivariate models exhibited the most-robust statistical power with more consistent results. These findings have important implications for metabolomics analysis in human disease.

The glucose-lowering effect of low-dose diacerein and its responsiveness metabolic markers in uncontrolled diabetes

Objective

Diacerein inhibits the synthesis and activity of pro-inflammatory cytokines, decreases macrophage infiltration in adipose tissue and thus increases insulin sensitivity and signalling. We conducted this study to determine the efficacy of low-dose diacerein in improving glycaemic control in type 2 diabetes mellitus (T2DM) patients with inadequate glycaemic control and to identify the metabolic determinants for such improvement. We randomised 25 T2DM patients with poor glycaemic control, despite being treated with at least three glucose-lowering agents, to receive diacerein 50 mg once-daily (n = 18) or placebo (n = 17) for 12 weeks. Changes in glycated haemoglobin (HbA1c) were evaluated at the 4th and 12th weeks. Metabolic profiling was performed using liquid chromatography electrospray ionisation quadrupole time-of-flight mass spectrometry.

Results

HbA1c levels were significantly reduced from baseline in the diacerein group at 12 weeks (− 0.6%, p < 0.05), whereas fasting plasma glucose (FPG) levels were not significantly decreased (− 18.9 mg/dl, p = 0.06). Partial least squares-discriminant analysis demonstrated an association between the serum abundance of threo-isocitric acid (ICA) and HbA1c response in the diacerein group. After adjusting for serum high-sensitivity C-reactive protein, ICA was still significantly related to the change in HbA1c.

Retrospective trial registration Current Controlled Trials TCTR20200820004, 20 August 2020.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13104-022-05974-9.

Contrasting signatures of genomic divergence during sympatric speciation

The transition from 'well-marked varieties' of a single species into 'well-defined species'-especially in the absence of geographic barriers to gene flow (sympatric speciation)-has puzzled evolutionary biologists ever since Darwin1,2. Gene flow counteracts the buildup of genome-wide differentiation, which is a hallmark of speciation and increases the likelihood of the evolution of irreversible reproductive barriers (incompatibilities) that complete the speciation process3. Theory predicts that the genetic architecture of divergently selected traits can influence whether sympatric speciation occurs4, but empirical tests of this theory are scant because comprehensive data are difficult to collect and synthesize across species, owing to their unique biologies and evolutionary histories5. Here, within a young species complex of neotropical cichlid fishes (Amphilophus spp.), we analysed genomic divergence among populations and species. By generating a new genome assembly and re-sequencing 453 genomes, we uncovered the genetic architecture of traits that have been suggested to be important for divergence. Species that differ in monogenic or oligogenic traits that affect ecological performance and/or mate choice show remarkably localized genomic differentiation. By contrast, differentiation among species that have diverged in polygenic traits is genomically widespread and much higher overall, consistent with the evolution of effective and stable genome-wide barriers to gene flow. Thus, we conclude that simple trait architectures are not always as conducive to speciation with gene flow as previously suggested, whereas polygenic architectures can promote rapid and stable speciation in sympatry.

Robust Wavelength Selection Using Filter-Wrapper Method and Input Scaling on Near Infrared Spectral Data

The extraction of relevant wavelengths from a large dataset of Near Infrared Spectroscopy (NIRS) is a significant challenge in vibrational spectroscopy research. Nonetheless, this process allows the improvement in the chemical interpretability by emphasizing the chemical entities related to the chemical parameters of samples. With the complexity in the dataset, it may be possible that irrelevant wavelengths are still included in the multivariate calibration. This yields the computational process to become unnecessary complex and decreases the accuracy and robustness of the model. In multivariate analysis, Partial Least Square Regression (PLSR) is a method commonly used to build a predictive model from NIR spectral data. However, in the PLSR method and common commercial chemometrics software, there is no standard wavelength selection procedure applied to screen the irrelevant wavelengths. In this study, a new robust wavelength selection procedure called the modified VIP-MCUVE (mod-VIP-MCUVE) using Filter-Wrapper method and input scaling strategy is introduced. The proposed method combines the modified Variable Importance in Projection (VIP) and modified Monte Carlo Uninformative Variable Elimination (MCUVE) to calculate the scale matrix of the input variable. The modified VIP uses the orthogonal components of Partial Least Square (PLS) in investigating the informative variable in the model by applying the amount of variation both in X and y{SSX,SSY}, simultaneously. The modified MCUVE uses a robust reliability coefficient and a robust tolerance interval in the selection procedure. To evaluate the superiority of the proposed method, the classical VIP, MCUVE, and autoscaling procedure in classical PLSR were also included in the evaluation. Using artificial data with Monte Carlo simulation and NIR spectral data of oil palm (Elaeis guineensis Jacq.) fruit mesocarp, the study shows that the proposed method offers advantages to improve model interpretability, to be computationally extensive, and to produce better model accuracy.

Input Selection Methods for Soft Sensor Design: A Survey

Soft Sensors (SSs) are inferential models used in many industrial fields. They allow for real-time estimation of hard-to-measure variables as a function of available data obtained from online sensors. SSs are generally built using industries historical databases through data-driven approaches. A critical issue in SS design concerns the selection of input variables, among those available in a candidate dataset. In the case of industrial processes, candidate inputs can reach great numbers, making the design computationally demanding and leading to poorly performing models. An input selection procedure is then necessary. Most used input selection approaches for SS design are addressed in this work and classified with their benefits and drawbacks to guide the designer through this step.

Independent and Opposite Associations Between Branched-Chain Amino Acids and Lysophosphatidylcholines With Incident Diabetes in Thais

Branched-chain amino acids (BCAAs) and lysophosphatidylcholines (LPCs) have been reported to be associated with diabetes. The purpose of the present study was to investigate the relative contributions of BCAAs and LPCs to the progression of prediabetes to diabetes using a targeted metabolomic approach. This study was part of a health survey of employees of the Electricity Generating Authority of Thailand (n = 79; nine females and 70 males). A targeted metabolomics analysis was performed using an AbsoluteIDQ^® p180 kit, flow injection analysis, and liquid chromatography-tandem mass spectrometry. The highest variable importance in projection (VIP) scores for the progression to diabetes of the amino acids and phospholipids were associated with isoleucine and LPC acyl C28:1, respectively. Using logistic regression analysis, we found that high baseline isoleucine concentration was associated with a higher incidence of diabetes, while high LPC acyl 28:1 was associated with a lower incidence. Isoleucine and LPC acyl 28:1 were independently associated with incident diabetes in a model that also included conventional risk factors for diabetes (baseline fasting plasma glucose (FPG), age, sex, and body mass index (BMI)). In addition, isoleucine and LPC acyl 28:1 were independently associated with serum HbA1c 5 years later in a robust regression model that also included baseline FPG, age, sex, and BMI. Isoleucine, LPC acyl 28:1, age, and FPG were significantly associated with HbA1c at this time. In conclusion, these results provide evidence that isoleucine and LPC acyl C28:1 have respective positive and negative independent associations with incident diabetes.

Cytotoxic comparison of macrolide antibiotics and their chlorinated disinfection byproduct mixtures

Erythromycin (ERY), azithromycin (AZI) and telithromycin (TEL) are widely-used macrolide antibiotics that are frequently detected in various water environments, including resource water and drinking water. In the performed chlorination disinfection process, at least 10, 20 and 200 new disinfection byproducts of ERY, AZI and TEL, respectively, were observed (the mixtures of the disinfection byproducts of ERY, AZI and TEL were named ERY-M, AZI-M and TEL-M, respectively). There is limited information available regarding their comparative toxicities, and their potential health risks are still unknown. In this study, the Jurkat cell line was used to compare the toxicities of the disinfection byproduct mixtures and their precursor compounds. The cell viability results indicated that the toxicity of ERY-M may not be enhanced after disinfection by chlorination. In contrast, at the same concentrations, AZI-M and TEL-M induced more significant inhibitory effects on cell viability than their parent compounds. Additionally, the total antioxidant capacity (T-AOC) and cell cytokine release (including interleukin-2, interleukin-8 and tumor necrosis factor-α) analyses of AZI-M and TEL-M further verified these results. Our findings demonstrate that the cytotoxicity of AZI and TEL was enhanced during the chlorination disinfection process. This investigation will provide substantial new details related to the toxicity of the mixed disinfection byproducts (DBPs) of ERY, AZI and TEL generated in the chlorination disinfection process.

Systematic Assessment of Immune Marker Variation in Type 1 Diabetes: A Prospective Longitudinal Study

Immune analytes have been widely tested in efforts to understand the heterogeneity of disease progression, risk, and therapeutic responses in type 1 diabetes (T1D). The future clinical utility of such analytes as biomarkers depends on their technical and biological variability, as well as their correlation with clinical outcomes. To assess the variability of a panel of 91 immune analytes, we conducted a prospective study of adults with T1D (<3 years from diagnosis), at 9–10 visits over 1 year. Autoantibodies and frequencies of T-cell, natural killer cell, and myeloid subsets were evaluated; autoreactive T-cell frequencies and function were also measured. We calculated an intraclass correlation coefficient (ICC) for each marker, which is a relative measure of between- and within-subject variability. Of the 91 analytes tested, we identified 35 with high between- and low within-subject variability, indicating their potential ability to be used to stratify subjects. We also provide extensive data regarding technical variability for 64 of the 91 analytes. To pilot the concept that ICC can be used to identify analytes that reflect biological outcomes, the association between each immune analyte and C-peptide was also evaluated using partial least squares modeling. CD8 effector memory T-cell (CD8 EM) frequency exhibited a high ICC and a positive correlation with C-peptide, which was also seen in an independent dataset of recent-onset T1D subjects. More work is needed to better understand the mechanisms underlying this relationship. Here we find that there are a limited number of technically reproducible immune analytes that also have a high ICC. We propose the use of ICC to define within- and between-subject variability and measurement of technical variability for future biomarker identification studies. Employing such a method is critical for selection of analytes to be tested in the context of future clinical trials aiming to understand heterogeneity in disease progression and response to therapy.

Statistical Workflow for Feature Selection in Human Metabolomics Data

High-throughput metabolomics investigations, when conducted in large human cohorts, represent a potentially powerful tool for elucidating the biochemical diversity underlying human health and disease. Large-scale metabolomics data sources, generated using either targeted or nontargeted platforms, are becoming more common. Appropriate statistical analysis of these complex high-dimensional data will be critical for extracting meaningful results from such large-scale human metabolomics studies. Therefore, we consider the statistical analytical approaches that have been employed in prior human metabolomics studies. Based on the lessons learned and collective experience to date in the field, we offer a step-by-step framework for pursuing statistical analyses of cohort-based human metabolomics data, with a focus on feature selection. We discuss the range of options and approaches that may be employed at each stage of data management, analysis, and interpretation and offer guidance on the analytical decisions that need to be considered over the course of implementing a data analysis workflow. Certain pervasive analytical challenges facing the field warrant ongoing focused research. Addressing these challenges, particularly those related to analyzing human metabolomics data, will allow for more standardization of as well as advances in how research in the field is practiced. In turn, such major analytical advances will lead to substantial improvements in the overall contributions of human metabolomics investigations.

Monitoring Radiotherapeutic Response in Prostate Cancer Patients Using High Throughput FTIR Spectroscopy of Liquid Biopsies

Radiation therapy (RT) is used to treat approximately 50% of all cancer patients. However, RT causes a wide range of adverse late effects that can affect a patient's quality of life. There are currently no predictive assays in clinical use to identify patients at risk of normal tissue radiation toxicity. This study aimed to investigate the potential of Fourier transform infrared (FTIR) spectroscopy for monitoring radiotherapeutic response. Blood plasma was acquired from 53 prostate cancer patients at five different time points: prior to treatment, after hormone treatment, at the end of radiotherapy, two months post radiotherapy and eight months post radiotherapy. FTIR spectra were recorded from plasma samples at all time points and the data was analysed using MATLAB software. Discrimination was observed between spectra recorded at baseline versus follow up time points, as well as between spectra from patients showing minimal and severe acute and late toxicity using principal component analysis. A partial least squares discriminant analysis model achieved sensitivity and specificity rates ranging from 80% to 99%. This technology may have potential to monitor radiotherapeutic response in prostate cancer patients using non-invasive blood plasma samples and could lead to individualised patient radiotherapy.

Variable selection and validation in multivariate modelling

MOTIVATION

Validation of variable selection and predictive performance is crucial in construction of robust multivariate models that generalize well, minimize overfitting and facilitate interpretation of results. Inappropriate variable selection leads instead to selection bias, thereby increasing the risk of model overfitting and false positive discoveries. Although several algorithms exist to identify a minimal set of most informative variables (i.e. the minimal-optimal problem), few can select all variables related to the research question (i.e. the all-relevant problem). Robust algorithms combining identification of both minimal-optimal and all-relevant variables with proper cross-validation are urgently needed.

RESULTS

We developed the MUVR algorithm to improve predictive performance and minimize overfitting and false positives in multivariate analysis. In the MUVR algorithm, minimal variable selection is achieved by performing recursive variable elimination in a repeated double cross-validation (rdCV) procedure. The algorithm supports partial least squares and random forest modelling, and simultaneously identifies minimal-optimal and all-relevant variable sets for regression, classification and multilevel analyses. Using three authentic omics datasets, MUVR yielded parsimonious models with minimal overfitting and improved model performance compared with state-of-the-art rdCV. Moreover, MUVR showed advantages over other variable selection algorithms, i.e. Boruta and VSURF, including simultaneous variable selection and validation scheme and wider applicability.

AVAILABILITY AND IMPLEMENTATION

Algorithms, data, scripts and tutorial are open source and available as an R package ('MUVR') at https://gitlab.com/CarlBrunius/MUVR.git.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Integrating management information with soil quality dynamics to monitor agricultural productivity

Sustainably utilizing global resources is critical for ensuring soil security which is pertinent for biomass production, climate change mitigation, environmental quality, biodiversity conservation and thus human wellbeing. A plethora of soil quality assessment metrics encapsulated in different concepts exist, with each typically biased towards identifying the interrelationship between agricultural production and specific physical, chemical or biological soil attributes. Because of diversity in soil classifications and crop requirements, considerable variation exist between these metrics making it difficult for end-users to select a suitable method. Here, Partial Least Squares Regression (PLSR) method is used to integrate the physical and chemical soil properties into a Soil Quality Index (SQI) which is then used to evaluate soil quality dynamics vis-à-vis crop yields over two growing seasons. Field data was acquired from 5 sites under No-Till (NT), Conventional Till (CT) management and Natural Vegetation (NV) land use. This SQI was computed under the hypothesis that site specific soil physico-chemical attributes depended on soil type, management, and depth. Under CT management Pw (Pewamo silty clay loam) had the highest soil quality; KbA (Kibbie fine sandy loam) soils had higher quality under NT management; whereas CtA (Crosby Celina silt loams) had relatively higher quality under NV land use. Soil bulk density (ρb), Soil Organic Carbon (SOC), Available Water Content (AWC) and Electrical Conductivity (EC) were the significant soil parameters influencing soil quality. The correlation between SQI and corn (Zea mays) yields was 0.6, whereas SQI and Soybean (Glycine max (L.) Merr.) yield was 0.9. Future research will evaluate SQI dynamics vis-à-vis socio-economic indicators and key climate variables.

An ensemble variable selection method for vibrational spectroscopic data analysis

Wavelength selection is a critical factor for pattern recognition of vibrational spectroscopic data. Not only does it alleviate the effect of dimensionality on an algorithm's generalization performance, but it also enhances the understanding and interpretability of multivariate classification models. In this study, a novel partial least squares discriminant analysis (PLSDA)-based wavelength selection algorithm, termed ensemble of bootstrapping space shrinkage (EBSS), has been devised for vibrational spectroscopic data analysis. In the algorithm, a set of subsets are generated from a data set using random sampling. For an individual subset, a feature space is determined by maximizing the expected 10-fold cross-validation accuracy with a weighted bootstrap sampling strategy. Then an ensemble strategy and a sequential forward selection method are applied to the feature spaces to select characteristic variables. Experimental results obtained from analysis of real vibrational spectroscopic data sets demonstrate that the ensemble wavelength selection algorithm can reserve stable and informative variables for the final modeling and improve predictive ability for multivariate classification models.

A new ensemble method for wavelength selection.

Partial Least Squares Discriminant Analysis and Bayesian Networks for Metabolomic Prediction of Childhood Asthma

To explore novel methods for the analysis of metabolomics data, we compared the ability of Partial Least Squares Discriminant Analysis (PLS-DA) and Bayesian networks (BN) to build predictive plasma metabolite models of age three asthma status in 411 three year olds (n = 59 cases and 352 controls) from the Vitamin D Antenatal Asthma Reduction Trial (VDAART) study. The standard PLS-DA approach had impressive accuracy for the prediction of age three asthma with an Area Under the Curve Convex Hull (AUCCH) of 81%. However, a permutation test indicated the possibility of overfitting. In contrast, a predictive Bayesian network including 42 metabolites had a significantly higher AUCCH of 92.1% (p for difference < 0.001), with no evidence that this accuracy was due to overfitting. Both models provided biologically informative insights into asthma; in particular, a role for dysregulated arginine metabolism and several exogenous metabolites that deserve further investigation as potential causative agents. As the BN model outperformed the PLS-DA model in both accuracy and decreased risk of overfitting, it may therefore represent a viable alternative to typical analytical approaches for the investigation of metabolomics data.

Synthesizing water quality indicators from standardized geospatial information to remedy water security challenges: A review

Water is vital not only for food, energy and sanitation but also for ecosystem functioning, human health, socio-economic progress and poverty reduction. Water security exists when all people have physical and economical access to sufficient, safe, and clean water that meets basic needs. However, water security is threatened by growing human population, episodic environmental disasters, indiscriminate land management practices, contaminants, and escalation in geopolitical conflicts. <3% of the estimated 1.4 billion cubic kilometers of water on earth is available for consumption. Although there exist a range of laboratory and field methods for measuring the chemical, physical and biological properties of water, the information available to the public remains inconsistent and patchy. To this end, we advance a new theory of a single-value objective water quality index (WQI) that considers the interaction between the above properties, to provide concise information for source water quality surveillance and monitoring. Although geospatial technologies such as remote sensing is credited as a high frequency spatiotemporal mapping tool, exiguous information is available on its application for constructing single-value WQIs. Besides, no remote sensing device exists that directly measures water quality, which must indirectly be inferred through modeling sensed remote sensing signals with measured water properties. This review not only highlights the water security conundrum but also provides an overview of methods for integrating geolocated qualitative (e.g., management data) with quantitative (i.e., measured water constituent properties) into a WQI.

Flavor and Acceptance of Roasted California Almonds During Accelerated Storage

Monitoring oxidative flavor changes in almonds is possible only if the chemical and sensory profile during roasting and storage is first established. Herein, almonds roasted at two different temperatures (115 and 152 °C) were stored at 39 °C for 0 to 12 months and were analyzed by headspace solid-phase microextraction gas chromatography-mass spectrometry, descriptive analysis, and consumer hedonic analysis. Volatile profiles, descriptive sensory profiles, and consumer hedonic scores were analyzed for predictive relationships. Descriptive attributes involving Roasted and Nutty as well as consumer liking were highest in fresh almonds, while flavors typically associated with oxidative rancidity such as Cardboard, Painty/Solvent, Soapy, and Total Oxidized increased during storage. Compounds most important for predicting rancidity-related attributes were lipid oxidation products, including pentanal, hexanal, heptanal, and octanal. Consumer liking was best predicted by similar compounds to those predicting Clean Nutty flavor, including Maillard reaction products such as 2- and 3-methylbutanal, 2-methylpyrazine, and 2,5-dimethylpyrazine.

Identification of Urinary Polyphenol Metabolite Patterns Associated with Polyphenol-Rich Food Intake in Adults from Four European Countries

We identified urinary polyphenol metabolite patterns by a novel algorithm that combines dimension reduction and variable selection methods to explain polyphenol-rich food intake, and compared their respective performance with that of single biomarkers in the European Prospective Investigation into Cancer and Nutrition (EPIC) study. The study included 475 adults from four European countries (Germany, France, Italy, and Greece). Dietary intakes were assessed with 24-h dietary recalls (24-HDR) and dietary questionnaires (DQ). Thirty-four polyphenols were measured by ultra-performance liquid chromatography-electrospray ionization-tandem mass spectrometry (UPLC-ESI-MS-MS) in 24-h urine. Reduced rank regression-based variable importance in projection (RRR-VIP) and least absolute shrinkage and selection operator (LASSO) methods were used to select polyphenol metabolites. Reduced rank regression (RRR) was then used to identify patterns in these metabolites, maximizing the explained variability in intake of pre-selected polyphenol-rich foods. The performance of RRR models was evaluated using internal cross-validation to control for over-optimistic findings from over-fitting. High performance was observed for explaining recent intake (24-HDR) of red wine (r = 0.65; AUC = 89.1%), coffee (r = 0.51; AUC = 89.1%), and olives (r = 0.35; AUC = 82.2%). These metabolite patterns performed better or equally well compared to single polyphenol biomarkers. Neither metabolite patterns nor single biomarkers performed well in explaining habitual intake (as reported in the DQ) of polyphenol-rich foods. This proposed strategy of biomarker pattern identification has the potential of expanding the currently still limited list of available dietary intake biomarkers.

Development of a robust and validated 2D-QSPR model for sweetness potency of diverse functional organic molecules

In the present report, we have developed a predictive QSPR model using only easily computable two-dimensional (2D) descriptors from diverse classes of sweetening agents to find out the key structural properties which regulate their sweet potency. The available data set was curated to remove salts, mixtures and compounds without having a definite structure. A k-fold double cross validation technique was employed for variable selection prior to development of the final model. The final model was developed using partial least squares (PLS) regression analysis and selected based on a mean absolute error (MAE) based criteria for the validation sets. The model was validated extensively using different internal and external validation strategies in accordance with the Organization for Economic Co-operation and Development (OECD) guidelines. This work presented development of a validated quantitative structure-property relationship (QSPR) model obtained from k-fold double cross-validation technique in order to find out the key structural information required to enhance the sweet potency of the molecules. Finally, we have designed and proposed 13 new molecules based on the insights obtained from the QSPR model. The designed compounds showed good in silico predicted sweetness potency with acceptable ADMET profile.

Identification of dietary patterns associated with obesity in a nationally representative survey of Canadian adults: application of a priori, hybrid, and simplified dietary pattern techniques

Background: Analyzing the effects of dietary patterns is an important approach for examining the complex role of nutrition in the etiology of obesity and chronic diseases.Objectives: The objectives of this study were to characterize the dietary patterns of Canadians with the use of a priori, hybrid, and simplified dietary pattern techniques, and to compare the associations of these patterns with obesity risk in individuals with and without chronic diseases (unhealthy and healthy obesity).Design: Dietary recalls from 11,748 participants (≥18 y of age) in the cross-sectional, nationally representative Canadian Community Health Survey 2.2 were used. A priori dietary pattern was characterized with the use of the previously validated 2015 Dietary Guidelines for Americans Adherence Index (DGAI). Weighted partial least squares (hybrid method) was used to derive an energy-dense (ED), high-fat (HF), low-fiber density (LFD) dietary pattern with the use of 38 food groups. The associations of derived dietary patterns with disease outcomes were then tested with the use of multinomial logistic regression.Results: An ED, HF, and LFD dietary pattern had high positive loadings for fast foods, carbonated drinks, and refined grains, and high negative loadings for whole fruits and vegetables (≥|0.17|). Food groups with a high loading were summed to form a simplified dietary pattern score. Moving from the first (healthiest) to the fourth (least healthy) quartiles of the ED, HF, and LFD pattern and the simplified dietary pattern scores was associated with increasingly elevated ORs for unhealthy obesity, with individuals in quartile 4 having an OR of 2.57 (95% CI: 1.75, 3.76) and 2.73 (95% CI: 1.88, 3.98), respectively (P-trend < 0.0001). Individuals who adhered the most to the 2015 DGAI recommendations (quartile 4) had a 53% lower OR of unhealthy obesity (P-trend < 0.0001). The associations of dietary patterns with healthy obesity and unhealthy nonobesity were weaker, albeit significant.Conclusions: Consuming an ED, HF, and LFD dietary pattern and lack of adherence to the recommendations of the 2015 DGAI were associated with a significantly higher risk of obesity with and without accompanying chronic diseases.

Unravelling the metabolic impact of SBS-associated microbial dysbiosis: Insights from the piglet short bowel syndrome model

Liver disease is a major source of morbidity and mortality in children with short bowel syndrome (SBS). SBS-associated microbial dysbiosis has recently been implicated in the development of SBS-associated liver disease (SBS-ALD), however the pathological implications of this association have not been explored. In this study high-throughput sequencing of colonic content from the well-validated piglet SBS-ALD model was examined to determine alterations in microbial communities, and concurrent metabolic alterations identified in urine samples via targeted mass spectrometry approaches (GC-MS, LC-MS, FIA-MS) further uncovered impacts of microbial disturbance on metabolic outcomes in SBS-ALD. Multi-variate analyses were performed to elucidate contributing SBS-ALD microbe and metabolite panels and to identify microbe-metabolite interactions. A unique SBS-ALD microbe panel was clearest at the genus level, with discriminating bacteria predominantly from the Firmicutes and Bacteroidetes phyla. The SBS-ALD metabolome included important alterations in the microbial metabolism of amino acids and the mitochondrial metabolism of branched chain amino acids. Correlation analysis defined microbe-metabolite clustering patterns unique to SBS-ALD and identified a metabolite panel that correlates with dysbiosis of the gut microbiome in SBS.

A standardized soil quality index for diverse field conditions

Understanding the nexus between soil quality and productivity is constrained by data artifacts, compounded by limitations of the existing models. Here, we explore the potential of 4 regression methods (i.e., Reduced Regression (RR), SIMPLS, Principal Component Regression (PCR), and Partial Least Squares Regression (PLSR)), to synthesize 10 soil physical and chemical properties acquired from 3 major management practices and different soil layers, into an unbiased soil quality index (SQI) capable of evaluating soil functions (e.g., biomass production). The data was acquired from privately owned fields within the state of Ohio, USA, at the following land use and management sites: natural vegetation (NV) or woodlands, conventional till (CT), and no-till (NT). The soils were sampled at similar landscape positions (i.e., summit) at depth intervals of 0-10, 10-20, 20-40 and 40-60 cm, and analyzed for bulk density (ρb), carbon/nitrogen (C/N) ratio, soil organic C (SOC), total N (TN), available water capacity (AWC), pH and electrical conductivity (EC). Preliminary analyses revealed the PLSR method as the most robust. The PLSR Variable Importance of Projection (VIP) was calculated, transformed into the SQI score and compared with yield data. SOC, ρb, C/N and EC were identified as the major variables influencing soil quality status. The data shows that the quality of Pewamo silty clay loam (Pw) soil was higher than Crosby Celina loams (CtA), Kibbie fine sandy loam (kbA), Glynwood silt loam (GWA) and Crosby silt loam (CrA), respectively. In 2012, the mean SQI was 42.9%, with corn and soybean yields of 7 and 2Mg/ha. The R(2) of SQI versus yield was 0.74 for corn (Zea mays L.), and 0.89 for soybean (Glycine max (L.) Merr.). Future studies will investigate techniques for mapping this SQI.

Fecal microbiota composition of breast-fed infants is correlated with human milk oligosaccharides consumed

OBJECTIVES

This study tested the hypothesis that the fecal bacterial genera of breast-fed (BF) and formula-fed (FF) infants differ and that human milk oligosaccharides (HMOs) modulate the microbiota of BF infants.

METHODS

Fecal samples were obtained from BF (n = 16) or FF (n = 6) infants at 3-month postpartum. Human milk samples were collected on the same day when feces were collected. The microbiota was assessed by pyrosequencing of bacterial 16S ribosomal RNA genes. HMOs were measured by high-performance liquid chromatography-chip time-of-flight mass spectrometry.

RESULTS

The overall microbiota of BF differed from that of FF (P = 0.005). Compared with FF, BF had higher relative abundances of Bacteroides, lower proportions of Clostridium XVIII, Lachnospiraceae incertae sedis, Streptococcus, Enterococcus, and Veillonella (P < 0.05). Bifidobacterium predominated in both BF and FF infants, with no difference in abundance between the 2 groups. The most abundant HMOs were lacto-N-tetraose + lacto-N-neotetraose (LNT + LNnT, 22.6%), followed by 2'-fucosyllactose (2'FL, 14.5%) and lacto-N-fucopentaose I (LNFP I, 9.5%). Partial least squares regression of HMO and microbiota showed several infant fecal bacterial genera could be predicted by their mothers' HMO profiles, and the important HMOs for the prediction of bacterial genera were identified by variable importance in the projection scores.

CONCLUSIONS

These results strengthen the established relation between HMO and the infant microbiota and identify statistical means whereby infant bacterial genera can be predicted by milk HMO. Future studies are needed to validate these findings and determine whether the supplementation of formula with defined HMO could selectively modify the gut microbiota.

A Partial Least Squares Based Procedure for Upstream Sequence Classification in Prokaryotes

The upstream region of coding genes is important for several reasons, for instance locating transcription factor, binding sites, and start site initiation in genomic DNA. Motivated by a recently conducted study, where multivariate approach was successfully applied to coding sequence modeling, we have introduced a partial least squares (PLS) based procedure for the classification of true upstream prokaryotic sequence from background upstream sequence. The upstream sequences of conserved coding genes over genomes were considered in analysis, where conserved coding genes were found by using pan-genomics concept for each considered prokaryotic species. PLS uses position specific scoring matrix (PSSM) to study the characteristics of upstream region. Results obtained by PLS based method were compared with Gini importance of random forest (RF) and support vector machine (SVM), which is much used method for sequence classification. The upstream sequence classification performance was evaluated by using cross validation, and suggested approach identifies prokaryotic upstream region significantly better to RF (p-value < 0.01) and SVM (p-value < 0.01). Further, the proposed method also produced results that concurred with known biological characteristics of the upstream region.

Association of a lifestyle index with MRI-determined liver fat content in a general population study

BACKGROUND

In prior studies, lifestyle indices were associated with numerous disease end points, but the association with fatty liver disease (FLD), a key correlate of cardiometabolic risk, is unknown. The aim was to investigate associations between a lifestyle index with liver fat content.

METHODS

Liver fat was quantified by MRI as liver signal intensity (LSI) in 354 individuals selected from a population-based cohort from Germany. Exposure to favourable lifestyle factors was quantified using an additive score with each factor modelled as a dichotomous trait. Favourable lifestyle factors were defined as waist circumference below 102 (men) or 88 cm (women), physical activity ≥3.5 h/week, never-smoking and a favourable dietary pattern, which was derived to explain liver fat variation. In a cross-sectional study, multivariable adjusted linear and logistic regression was applied to investigate the association between the lifestyle index (range 0-4, exposure) and LSI (modelled as a continuous trait or dichotomised as a FLD indicator variable, respectively).

RESULTS

Individuals with four favourable lifestyle factors (n=9%) had lower LSI values (ß -0.40; 95% CI -0.61 to -0.19) and a lower OR (0.09; 95% CI 0.03 to 0.30) for FLD compared with individuals with zero favourable lifestyle factors (n=10%).

CONCLUSIONS

A healthy lifestyle pattern was associated with less liver fat. Prospective studies are warranted.

Monitoring cellular stress responses using integrated high-frequency impedance spectroscopy and time-resolved ELISA

We have developed a lab-on-a-chip system for continuous and non-invasive monitoring of microfluidic cell cultures using integrated high-frequency contactless impedance spectroscopy. Electrically insulated microfabricated interdigitated electrode structures were embedded into four individually addressable microchambers to reliably and reproducibly detect cell-substrate interactions, cell viability and metabolic activity. While silicon nitride passivated sensor substrates provided a homogeneous cell culture surface that minimized cell orientation along interdigitated electrode structures, the application of high-frequency AC fields reduced the impact of the 300 nm thick passivation layer on sensor sensitivity. The additional implementation of multivariate data analysis methods such as partial least square (PLS) for high-frequency impedance spectra provided unambiguous information on intracellular pathway activation, up and down-regulation of protein synthesis as well as global cellular stress responses. A comparative cell analysis using connective tissue fibroblasts showed that high-frequency contactless impedance spectroscopy and time-resolved quantification of IL-6 secretion using ELISA provided similar results following stimulation with circulating pro-inflammatory cytokines IL-1β and TNFα. The combination of microfluidics with contactless impedance sensing and time-resolved quantification of stress factor release will provide biologist with a new tool to (a) establish a variety of uniform cell culture surfaces that feature complex biochemistries, micro- and nanopatterns; and (b) to simultaneously characterize cell responses under physiologically relevant conditions using a complementary non-invasive cell analysis method.

Dietary patterns associated with magnetic resonance imaging-determined liver fat content in a general population study

BACKGROUND

The association between diet and fatty liver disease (FLD) has predominantly been analyzed for single nutrients or foods, and findings have been inconsistent.

OBJECTIVE

We aimed to compare associations of hypothesis-driven and exploratory dietary pattern scores with liver fat content.

DESIGN

Liver fat was measured by using magnetic resonance imaging as liver signal intensity (LSI) in a population-based, cross-sectional study that included 354 individuals. We applied partial least-squares regression to derive an exploratory dietary pattern score that explained variation in both the intake of 38 food groups, which were assessed by using a food-frequency questionnaire, and LSI. The hypothesis-driven score was calculated on the basis of published studies. Multivariable linear or logistic regression was used to investigate associations between dietary pattern scores and LSI or FLD.

RESULTS

A higher percentage of LSI variation was explained by the exploratory (12.6%) compared with the hypothesis-driven (2.2%) dietary pattern. Of the 13 most important food groups of the exploratory dietary pattern, intakes of green and black tea, soups, and beer were also individually associated with LSI values. A 1-unit increase in the exploratory dietary pattern score was positively associated with FLD (OR: 1.56; 95% CI: 1.29, 1.88). Furthermore, a 1-unit increase in the hypothesis-driven dietary pattern score, which consisted of alcohol, soft drinks, meat, coffee, and tea, was positively associated with FLD (OR: 1.25; 95% CI: 1.10, 1.43).

CONCLUSION

We defined a hypothesis-driven dietary pattern and derived an exploratory dietary pattern, both of which included alcohol, meat (poultry), and tea, associated with liver fat content independent from confounders, which should be explored in prospective studies.

Improving stability and understandability of genotype-phenotype mapping in Saccharomyces using regularized variable selection in L-PLS regression

Background

Multivariate approaches have been successfully applied to genome wide association studies. Recently, a Partial Least Squares (PLS) based approach was introduced for mapping yeast genotype-phenotype relations, where background information such as gene function classification, gene dispensability, recent or ancient gene copy number variations and the presence of premature stop codons or frameshift mutations in reading frames, were used post hoc to explain selected genes. One of the latest advancement in PLS named L-Partial Least Squares (L-PLS), where ‘L’ presents the used data structure, enables the use of background information at the modeling level. Here, a modification of L-PLS with variable importance on projection (VIP) was implemented using a stepwise regularized procedure for gene and background information selection. Results were compared to PLS-based procedures, where no background information was used.

Results

Applying the proposed methodology to yeast Saccharomyces cerevisiae data, we found the relationship between genotype-phenotype to have improved understandability. Phenotypic variations were explained by the variations of relatively stable genes and stable background variations. The suggested procedure provides an automatic way for genotype-phenotype mapping. The selected phenotype influencing genes were evolving 29% faster than non-influential genes, and the current results are supported by a recently conducted study. Further power analysis on simulated data verified that the proposed methodology selects relevant variables.

Conclusions

A modification of L-PLS with VIP in a stepwise regularized elimination procedure can improve the understandability and stability of selected genes and background information. The approach is recommended for genome wide association studies where background information is available.