A network-based feature selection approach to identify metabolic signatures in disease

Network-based artificial intelligence approaches for advancing personalized psychiatry

Psychiatric disorders have a complex biological underpinning likely involving an interplay of genetic and environmental risk contributions. Substantial efforts are being made to use artificial intelligence approaches to integrate features within and across data types to increase our etiological understanding and advance personalized psychiatry. Network science offers a conceptual framework for exploring the often complex relationships across different levels of biological organization, from cellular mechanistic to brain-functional and phenotypic networks. Utilizing such network information effectively as part of artificial intelligence approaches is a promising route toward a more in-depth understanding of illness biology, the deciphering of patient heterogeneity, and the identification of signatures that may be sufficiently predictive to be clinically useful. Here, we present examples of how network information has been used as part of artificial intelligence within psychiatry and beyond and outline future perspectives on how personalized psychiatry approaches may profit from a closer integration of psychiatric research, artificial intelligence development, and network science.

Multi-scale supervised clustering-based feature selection for tumor classification and identification of biomarkers and targets on genomic data

Background

The small number of samples and the curse of dimensionality hamper the better application of deep learning techniques for disease classification. Additionally, the performance of clustering-based feature selection algorithms is still far from being satisfactory due to their limitation in using unsupervised learning methods. To enhance interpretability and overcome this problem, we developed a novel feature selection algorithm. In the meantime, complex genomic data brought great challenges for the identification of biomarkers and therapeutic targets. The current some feature selection methods have the problem of low sensitivity and specificity in this field.

Results

In this article, we designed a multi-scale clustering-based feature selection algorithm named MCBFS which simultaneously performs feature selection and model learning for genomic data analysis. The experimental results demonstrated that MCBFS is robust and effective by comparing it with seven benchmark and six state-of-the-art supervised methods on eight data sets. The visualization results and the statistical test showed that MCBFS can capture the informative genes and improve the interpretability and visualization of tumor gene expression and single-cell sequencing data. Additionally, we developed a general framework named McbfsNW using gene expression data and protein interaction data to identify robust biomarkers and therapeutic targets for diagnosis and therapy of diseases. The framework incorporates the MCBFS algorithm, network recognition ensemble algorithm and feature selection wrapper. McbfsNW has been applied to the lung adenocarcinoma (LUAD) data sets. The preliminary results demonstrated that higher prediction results can be attained by identified biomarkers on the independent LUAD data set, and we also structured a drug-target network which may be good for LUAD therapy.

Conclusions

The proposed novel feature selection method is robust and effective for gene selection, classification, and visualization. The framework McbfsNW is practical and helpful for the identification of biomarkers and targets on genomic data. It is believed that the same methods and principles are extensible and applicable to other different kinds of data sets.

A protein network descriptor server and its use in studying protein, disease, metabolic and drug targeted networks

The genetic, proteomic, disease and pharmacological studies have generated rich data in protein interaction, disease regulation and drug activities useful for systems-level study of the biological, disease and drug therapeutic processes. These studies are facilitated by the established and the emerging computational methods. More recently, the network descriptors developed in other disciplines have become more increasingly used for studying the protein-protein, gene regulation, metabolic, disease networks. There is an inadequate coverage of these useful network features in the public web servers. We therefore introduced upto 313 literature-reported network descriptors in PROFEAT web server, for describing the topological, connectivity and complexity characteristics of undirected unweighted (uniform binding constants and molecular levels), undirected edge-weighted (varying binding constants), undirected node-weighted (varying molecular levels), undirected edge-node-weighted (varying binding constants and molecular levels) and directed unweighted (oriented process) networks. The usefulness of the PROFEAT computed network descriptors is illustrated by their literature-reported applications in studying the protein-protein, gene regulatory, gene co-expression, protein-drug and metabolic networks. PROFEAT is accessible free of charge at http://bidd2.nus.edu.sg/cgi-bin/profeat2016/main.cgi.

A New Strategy for Analyzing Time-Series Data Using Dynamic Networks: Identifying Prospective Biomarkers of Hepatocellular Carcinoma

Time-series metabolomics studies can provide insight into the dynamics of disease development and facilitate the discovery of prospective biomarkers. To improve the performance of early risk identification, a new strategy for analyzing time-series data based on dynamic networks (ATSD-DN) in a systematic time dimension is proposed. In ATSD-DN, the non-overlapping ratio was applied to measure the changes in feature ratios during the process of disease development and to construct dynamic networks. Dynamic concentration analysis and network topological structure analysis were performed to extract early warning information. This strategy was applied to the study of time-series lipidomics data from a stepwise hepatocarcinogenesis rat model. A ratio of lyso-phosphatidylcholine (LPC) 18:1/free fatty acid (FFA) 20:5 was identified as the potential biomarker for hepatocellular carcinoma (HCC). It can be used to classify HCC and non-HCC rats, and the area under the curve values in the discovery and external validation sets were 0.980 and 0.972, respectively. This strategy was also compared with a weighted relative difference accumulation algorithm (wRDA), multivariate empirical Bayes statistics (MEBA) and support vector machine-recursive feature elimination (SVM-RFE). The better performance of ATSD-DN suggests its potential for a more complete presentation of time-series changes and effective extraction of early warning information.

Modeling and Classification of Kinetic Patterns of Dynamic Metabolic Biomarkers in Physical Activity

The objectives of this work were the classification of dynamic metabolic biomarker candidates and the modeling and characterization of kinetic regulatory mechanisms in human metabolism with response to external perturbations by physical activity. Longitudinal metabolic concentration data of 47 individuals from 4 different groups were examined, obtained from a cycle ergometry cohort study. In total, 110 metabolites (within the classes of acylcarnitines, amino acids, and sugars) were measured through a targeted metabolomics approach, combining tandem mass spectrometry (MS/MS) with the concept of stable isotope dilution (SID) for metabolite quantitation. Biomarker candidates were selected by combined analysis of maximum fold changes (MFCs) in concentrations and P-values resulting from statistical hypothesis testing. Characteristic kinetic signatures were identified through a mathematical modeling approach utilizing polynomial fitting. Modeled kinetic signatures were analyzed for groups with similar behavior by applying hierarchical cluster analysis. Kinetic shape templates were characterized, defining different forms of basic kinetic response patterns, such as sustained, early, late, and other forms, that can be used for metabolite classification. Acetylcarnitine (C2), showing a late response pattern and having the highest values in MFC and statistical significance, was classified as late marker and ranked as strong predictor (MFC = 1.97, P < 0.001). In the class of amino acids, highest values were shown for alanine (MFC = 1.42, P < 0.001), classified as late marker and strong predictor. Glucose yields a delayed response pattern, similar to a hockey stick function, being classified as delayed marker and ranked as moderate predictor (MFC = 1.32, P < 0.001). These findings coincide with existing knowledge on central metabolic pathways affected in exercise physiology, such as β-oxidation of fatty acids, glycolysis, and glycogenolysis. The presented modeling approach demonstrates high potential for dynamic biomarker identification and the investigation of kinetic mechanisms in disease or pharmacodynamics studies using MS data from longitudinal cohort studies.

Human metabolism is controlled through basic kinetic regulatory mechanisms, where the overall system aims to maintain a state of homeostasis. In response to external perturbations, such as environmental influences, nutrition or physical exercise, circulating metabolites show specific kinetic response patterns, which can be computationally modeled. In this work, we searched for dynamic metabolic biomarker candidates and analyzed specific kinetic mechanisms from longitudinal metabolic concentration data, obtained through a cycle ergometry stress test. In total, 110 metabolites measured from blood samples of 47 individuals were analyzed using tandem mass spectrometry (MS/MS). Dynamic biomarker candidates could be selected based on the amplitudes of changes in metabolite concentrations and the significance of statistical hypothesis testing. We were able to characterize specific kinetic patterns for groups of similarly behaving metabolites. Kinetic shape templates were identified, defining basic kinetic response patterns to physical exercise, such as sustained, early, late and other shape forms. The presented approach contributes to a better understanding of (patho)physiological biochemical mechanisms in human health, disease or during drug therapy, by offering tools for classifying dynamic biomarker candidates and for modeling and characterizing kinetic regulatory mechanisms from longitudinal experimental data.

Metabolic biomarkers for chronic kidney disease

Chronic kidney disease (CKD) is an increasingly recognized burden for patients and health care systems with high (and growing) global incidence and prevalence, significant mortality, and disproportionately high treatment costs. Yet, the available diagnostic tools are either impractical in clinical routine or have serious shortcomings impeding a well-informed disease management although optimized treatment strategies with proven benefits for the patients have become available. Advances in bioanalytical technologies have facilitated studies that identified genomic, proteomic, and metabolic biomarker candidates, and confirmed some of them in independent cohorts. This review summarizes the CKD-related markers discovered so far, and focuses on compounds and pathways, for which there is quantitative data, substantiating evidence from translational research, and a mechanistic understanding of the processes involved. Also, multiparametric marker panels have been suggested that showed promising diagnostic and prognostic performance in initial analyses although the data basis from prospective trials is very limited. Large-scale studies, however, are underway and will provide the information for validating a set of parameters and discarding others. Finally, the path from clinical research to a routine application is discussed, focusing on potential obstacles such as the use of mass spectrometry, and the feasibility of obtaining regulatory approval for targeted metabolomics assays.

Analytical methods in untargeted metabolomics: state of the art in 2015

Metabolomics comprises the methods and techniques that are used to measure the small molecule composition of biofluids and tissues, and is actually one of the most rapidly evolving research fields. The determination of the metabolomic profile - the metabolome - has multiple applications in many biological sciences, including the developing of new diagnostic tools in medicine. Recent technological advances in nuclear magnetic resonance and mass spectrometry are significantly improving our capacity to obtain more data from each biological sample. Consequently, there is a need for fast and accurate statistical and bioinformatic tools that can deal with the complexity and volume of the data generated in metabolomic studies. In this review, we provide an update of the most commonly used analytical methods in metabolomics, starting from raw data processing and ending with pathway analysis and biomarker identification. Finally, the integration of metabolomic profiles with molecular data from other high-throughput biotechnologies is also reviewed.

Metabolomic biomarkers for obesity in humans: a short review

BACKGROUND

The prevalence and incidence of obesity have become a major public health problem during the last decades, but the underlying biochemical and metabolic processes are not fully understood. Metabolomics, the science of small molecules of the metabolism, is helping to unravel these mechanisms via the identification of markers related to obesity. These biomarkers are used to prevent diseases in later life or for the early diagnosis of diseases. This review focuses on articles dealing with biomarkers for obesity.

KEY MESSAGES

Branched-chain amino acids (BCAA), nonesterified fatty acids, organic acids, acylcarnitines, and phospholipids were identified as potential biomarkers for obesity. This indicates a relation between elevated BCAA, and other amino acids, and the obese state. Furthermore, deregulation of β-oxidation is associated with the development of obesity. The results have several limitations, including the differing ages of the subjects in the studies, the fact that all of the studies had a case-control design and therefore no causal explanatory power, and that most looked for similar metabolites and reported almost equal results.

CONCLUSION

The strength of this review is that it gives a comprehensive overview of the current status of the knowledge on metabolomics biomarkers for obesity, but further research is needed because the methods used in the studies to date are very homogenous, e.g. most used a targeted approach and therefore analyzed almost the same group of metabolites. Moreover, prospective studies are lacking since all of the studies are either case-control or cross-sectional studies.

Analyzing methods for path mining with applications in metabolomics

Metabolomics is one of the key approaches of systems biology that consists of studying biochemical networks having a set of metabolites, enzymes, reactions and their interactions. As biological networks are very complex in nature, proper techniques and models need to be chosen for their better understanding and interpretation. One of the useful strategies in this regard is using path mining strategies and graph-theoretical approaches that help in building hypothetical models and perform quantitative analysis. Furthermore, they also contribute to analyzing topological parameters in metabolome networks. Path mining techniques can be based on grammars, keys, patterns and indexing. Moreover, they can also be used for modeling metabolome networks, finding structural similarities between metabolites, in-silico metabolic engineering, shortest path estimation and for various graph-based analysis. In this manuscript, we have highlighted some core and applied areas of path-mining for modeling and analysis of metabolic networks.