Application of ensemble deep neural network to metabolomics studies

Transcriptome analysis of the Japanese eel (Anguilla japonica) during larval metamorphosis

BACKGROUND

Anguillid eels spend their larval period as leptocephalus larvae that have a unique and specialized body form with leaf-like and transparent features, and they undergo drastic metamorphosis to juvenile glass eels. Less is known about the transition of leptocephali to the glass eel stage, because it is difficult to catch the metamorphosing larvae in the open ocean. However, recent advances in rearing techniques for the Japanese eel have made it possible to study the larval metamorphosis of anguillid eels. In the present study, we investigated the dynamics of gene expression during the metamorphosis of Japanese eel leptocephali using RNA sequencing.

RESULTS

During metamorphosis, Japanese eels were classified into 7 developmental stages according to their morphological characteristics, and RNA sequencing was used to collect gene expression data from each stage. A total of 354.8 million clean reads were generated from the body and 365.5 million from the head, after the processing of raw reads. For filtering of genes that characterize developmental stages, a classification model created by a Random Forest algorithm was built. Using the importance of explanatory variables feature obtained from the created model, we identified 46 genes selected in the body and 169 genes selected in the head that were defined as the "most characteristic genes" during eel metamorphosis. Next, network analysis and subsequently gene clustering were conducted using the most characteristic genes and their correlated genes, and then 6 clusters in the body and 5 clusters in the head were constructed. Then, the characteristics of the clusters were revealed by Gene Ontology (GO) enrichment analysis. The expression patterns and GO terms of each stage were consistent with previous observations and experiments during the larval metamorphosis of the Japanese eel.

CONCLUSION

Genome and transcriptome resources have been generated for metamorphosing Japanese eels. Genes that characterized metamorphosis of the Japanese eel were identified through statistical modeling by a Random Forest algorithm. The functions of these genes were consistent with previous observations and experiments during the metamorphosis of anguillid eels.

How is Big Data reshaping preclinical aging research?

The exponential scientific and technological progress during the past 30 years has favored the comprehensive characterization of aging processes with their multivariate nature, leading to the advent of Big Data in preclinical aging research. Spanning from molecular omics to organism-level deep phenotyping, Big Data demands large computational resources for storage and analysis, as well as new analytical tools and conceptual frameworks to gain novel insights leading to discovery. Systems biology has emerged as a paradigm that utilizes Big Data to gain insightful information enabling a better understanding of living organisms, visualized as multilayered networks of interacting molecules, cells, tissues and organs at different spatiotemporal scales. In this framework, where aging, health and disease represent emergent states from an evolving dynamic complex system, context given by, for example, strain, sex and feeding times, becomes paramount for defining the biological trajectory of an organism. Using bioinformatics and artificial intelligence, the systems biology approach is leading to remarkable advances in our understanding of the underlying mechanism of aging biology and assisting in creative experimental study designs in animal models. Future in-depth knowledge acquisition will depend on the ability to fully integrate information from different spatiotemporal scales in organisms, which will probably require the adoption of theories and methods from the field of complex systems. Here we review state-of-the-art approaches in preclinical research, with a focus on rodent models, that are leading to conceptual and/or technical advances in leveraging Big Data to understand basic aging biology and its full translational potential.

Opportunities and challenges for sample preparation and enrichment in mass spectrometry for single-cell metabolomics

Single-cell heterogeneity in metabolism, drug resistance and disease type poses the need for analytical techniques for single-cell analysis. As the metabolome provides the closest view of the status quo in the cell, studying the metabolome at single-cell resolution may unravel said heterogeneity. A challenge in single-cell metabolome analysis is that metabolites cannot be amplified, so one needs to deal with picolitre volumes and a wide range of analyte concentrations. Due to high sensitivity and resolution, MS is preferred in single-cell metabolomics. Large numbers of cells need to be analysed for proper statistics; this requires high-throughput analysis, and hence automation of the analytical workflow. Significant advances in (micro)sampling methods, CE and ion mobility spectrometry have been made, some of which have been applied in high-throughput analyses. Microfluidics has enabled an automation of cell picking and metabolite extraction; image recognition has enabled automated cell identification. Many techniques have been used for data analysis, varying from conventional techniques to novel combinations of advanced chemometric approaches. Steps have been set in making data more findable, accessible, interoperable and reusable, but significant opportunities for improvement remain. Herein, advances in single-cell analysis workflows and data analysis are discussed, and recommendations are made based on the experimental goal.

Accurate Classification and Prediction of Acute Myocardial Infarction through an ARMD Procedure

The risk stratification of acute myocardial infarction (AMI) patients is of prime importance for clinical management and prognosis assessment. Thus, we propose an ensemble machine learning analysis procedure named ADASYN-RFECV-MDA-DNN (ARMD) to address sample-unbalanced problems and enable stratification and prediction of AMI outcomes. The ARMD analysis procedure was applied to the NMR data of sera from 534 AMI-related subjects in four categories with an extremely imbalanced sample proportion. Firstly, the adaptive synthetic sampling (ADASYN) algorithm was used to address the issue of the original sample imbalance. Secondly, the recursive feature elimination with cross-validation (RFECV) processing and random forest mean decrease accuracy (RF-MDA) algorithm was performed to identify the differential metabolites corresponding to each AMI outcome. Finally, the deep neural network (DNN) was employed to classify and predict AMI events, and its performance was evaluated by comparing the four traditional machine learning methods. Compared with the other four machine learning models, DNN presented consistent superiority in almost all of the model parameters including precision, f1-score, sensitivity, specificity, area under the receiver operating characteristic curve (AUC), and classification accuracy, highlighting the potential of deep learning in classification and stratification of clinical diseases. The ARMD analysis procedure was a practical analysis tool for supervised classification and regression modeling of clinical diseases.

An evaluation of homeostatic plasticity for ecosystems using an analytical data science approach

The natural world is constantly changing, and planetary boundaries are issuing severe warnings about biodiversity and cycles of carbon, nitrogen, and phosphorus. In other views, social problems such as global warming and food shortages are spreading to various fields. These seemingly unrelated issues are closely related, but it can be said that understanding them in an integrated manner is still a step away. However, progress in analytical technologies has been recognized in various fields and, from a microscopic perspective, with the development of instruments including next-generation sequencers (NGS), nuclear magnetic resonance (NMR), gas chromatography-mass spectrometry (GC/MS), and liquid chromatography-mass spectrometry (LC/MS), various forms of molecular information such as genome data, microflora structure, metabolome, proteome, and lipidome can be obtained. The development of new technology has made it possible to obtain molecular information in a variety of forms. From a macroscopic perspective, the development of environmental analytical instruments and environmental measurement facilities such as satellites, drones, observation ships, and semiconductor censors has increased the data availability for various environmental factors. Based on these background, the role of computational science is to provide a mechanism for integrating and understanding these seemingly disparate data sets. This review describes machine learning and the need for structural equations and statistical causal inference of these data to solve these problems. In addition to introducing actual examples of how these technologies can be utilized, we will discuss how to use these technologies to implement environmentally friendly technologies in society.

Interpretable machine learning methods for predictions in systems biology from omics data

Machine learning has become a powerful tool for systems biologists, from diagnosing cancer to optimizing kinetic models and predicting the state, growth dynamics, or type of a cell. Potential predictions from complex biological data sets obtained by “omics” experiments seem endless, but are often not the main objective of biological research. Often we want to understand the molecular mechanisms of a disease to develop new therapies, or we need to justify a crucial decision that is derived from a prediction. In order to gain such knowledge from data, machine learning models need to be extended. A recent trend to achieve this is to design “interpretable” models. However, the notions around interpretability are sometimes ambiguous, and a universal recipe for building well-interpretable models is missing. With this work, we want to familiarize systems biologists with the concept of model interpretability in machine learning. We consider data sets, data preparation, machine learning methods, and software tools relevant to omics research in systems biology. Finally, we try to answer the question: “What is interpretability?” We introduce views from the interpretable machine learning community and propose a scheme for categorizing studies on omics data. We then apply these tools to review and categorize recent studies where predictive machine learning models have been constructed from non-sequential omics data.

Omics Data and Data Representations for Deep Learning-Based Predictive Modeling

Medical discoveries mainly depend on the capability to process and analyze biological datasets, which inundate the scientific community and are still expanding as the cost of next-generation sequencing technologies is decreasing. Deep learning (DL) is a viable method to exploit this massive data stream since it has advanced quickly with there being successive innovations. However, an obstacle to scientific progress emerges: the difficulty of applying DL to biology, and this because both fields are evolving at a breakneck pace, thus making it hard for an individual to occupy the front lines of both of them. This paper aims to bridge the gap and help computer scientists bring their valuable expertise into the life sciences. This work provides an overview of the most common types of biological data and data representations that are used to train DL models, with additional information on the models themselves and the various tasks that are being tackled. This is the essential information a DL expert with no background in biology needs in order to participate in DL-based research projects in biomedicine, biotechnology, and drug discovery. Alternatively, this study could be also useful to researchers in biology to understand and utilize the power of DL to gain better insights into and extract important information from the omics data.

Learning a confidence score and the latent space of a new supervised autoencoder for diagnosis and prognosis in clinical metabolomic studies

Background

Presently, there is a wide variety of classification methods and deep neural network approaches in bioinformatics. Deep neural networks have proven their effectiveness for classification tasks, and have outperformed classical methods, but they suffer from a lack of interpretability. Therefore, these innovative methods are not appropriate for decision support systems in healthcare. Indeed, to allow clinicians to make informed and well thought out decisions, the algorithm should provide the main pieces of information used to compute the predicted diagnosis and/or prognosis, as well as a confidence score for this prediction.

Methods

Herein, we used a new supervised autoencoder (SAE) approach for classification of clinical metabolomic data. This new method has the advantage of providing a confidence score for each prediction thanks to a softmax classifier and a meaningful latent space visualization and to include a new efficient feature selection method, with a structured constraint, which allows for biologically interpretable results.

Results

Experimental results on three metabolomics datasets of clinical samples illustrate the effectiveness of our SAE and its confidence score. The supervised autoencoder provides an accurate localization of the patients in the latent space, and an efficient confidence score. Experiments show that the SAE outperforms classical methods (PLS-DA, Random Forests, SVM, and neural networks (NN)). Furthermore, the metabolites selected by the SAE were found to be biologically relevant.

Conclusion

In this paper, we describe a new efficient SAE method to support diagnostic or prognostic evaluation based on metabolomics analyses.

Parameter Visualization of Benchtop Nuclear Magnetic Resonance Spectra toward Food Process Monitoring

Low-cost and user-friendly benchtop low-field nuclear magnetic resonance (NMR) spectrometers are typically used to monitor food processes in the food industry. Because of excessive spectral overlap, it is difficult to characterize food mixtures using low-field NMR spectroscopy. In addition, for standard compounds, low-field benchtop NMR data are typically unavailable compared to high-field NMR data, which have been accumulated and are reusable in public databases. This work focused on NMR parameter visualization of the chemical structure and mobility of mixtures and the use of high-field NMR data to analyze benchtop NMR data to characterize food process samples. We developed a tool to easily process benchtop NMR data and obtain chemical shifts and T2 relaxation times of peaks, as well as transform high-field NMR data into low-field NMR data. Line broadening and time–frequency analysis methods were adopted for data processing. This tool can visualize NMR parameters to characterize changes in the components and mobilities of food process samples using benchtop NMR data. In addition, assignment errors were smaller when the spectra of standard compounds were identified by transferring the high-field NMR data to low-field NMR data rather than directly using experimentally obtained low-field NMR spectra.

Deep regression with ensembles enables fast, first-order shimming in low-field NMR

Shimming in the context of nuclear magnetic resonance aims to achieve a uniform magnetic field distribution, as perfect as possible, and is crucial for useful spectroscopy and imaging. Currently, shimming precedes most acquisition procedures in the laboratory, and this mostly semi-automatic procedure often needs to be repeated, which can be cumbersome and time-consuming. The paper investigates the feasibility of completely automating and accelerating the shimming procedure by applying deep learning (DL). We show that DL can relate measured spectral shape to shim current specifications and thus rapidly predict three shim currents simultaneously, given only four input spectra. Due to the lack of accessible data for developing shimming algorithms, we also introduce a database that served as our DL training set, and allows inference of changes to ¹H NMR signals depending on shim offsets. In situ experiments of deep regression with ensembles demonstrate a high success rate in spectral quality improvement for random shim distortions over different neural architectures and chemical substances. This paper presents a proof-of-concept that machine learning can simplify and accelerate the shimming problem, either as a stand-alone method, or in combination with traditional shimming methods. Our database and code are publicly available.

Multivariate analysis of NMR-based metabolomic data

Nuclear magnetic resonance (NMR) spectroscopy allows for simultaneous detection of a wide range of metabolites and lipids. As metabolites act together in complex metabolic networks, they are often highly correlated, and optimal biological insight is achieved when using methods that take the correlation into account. For this reason, latent-variable-based methods, such as principal component analysis and partial least-squares discriminant analysis, are widely used in metabolomic studies. However, with increasing availability of larger population cohorts, and a shift from analysis of spectral data to using quantified metabolite levels, both more traditional statistical approaches and alternative machine learning methods have become more widely used. This review aims at providing an overview of the current state-of-the-art multivariate methods for the analysis of NMR-based metabolomic data as well as alternative methods, highlighting their strengths and limitations.

Machine learning meets omics: applications and perspectives

The innovation of biotechnologies has allowed the accumulation of omics data at an alarming rate, thus introducing the era of 'big data'. Extracting inherent valuable knowledge from various omics data remains a daunting problem in bioinformatics. Better solutions often need some kind of more innovative methods for efficient handlings and effective results. Recent advancements in integrated analysis and computational modeling of multi-omics data helped address such needs in an increasingly harmonious manner. The development and application of machine learning have largely advanced our insights into biology and biomedicine and greatly promoted the development of therapeutic strategies, especially for precision medicine. Here, we propose a comprehensive survey and discussion on what happened, is happening and will happen when machine learning meets omics. Specifically, we describe how artificial intelligence can be applied to omics studies and review recent advancements at the interface between machine learning and the ever-widest range of omics including genomics, transcriptomics, proteomics, metabolomics, radiomics, as well as those at the single-cell resolution. We also discuss and provide a synthesis of ideas, new insights, current challenges and perspectives of machine learning in omics.

The exposome paradigm to predict environmental health in terms of systemic homeostasis and resource balance based on NMR data science

The environment, from microbial ecosystems to recycled resources, fluctuates dynamically due to many physical, chemical and biological factors, the profile of which reflects changes in overall state, such as environmental illness caused by a collapse of homeostasis. To evaluate and predict environmental health in terms of systemic homeostasis and resource balance, a comprehensive understanding of these factors requires an approach based on the "exposome paradigm", namely the totality of exposure to all substances. Furthermore, in considering sustainable development to meet global population growth, it is important to gain an understanding of both the circulation of biological resources and waste recycling in human society. From this perspective, natural environment, agriculture, aquaculture, wastewater treatment in industry, biomass degradation and biodegradable materials design are at the forefront of current research. In this respect, nuclear magnetic resonance (NMR) offers tremendous advantages in the analysis of samples of molecular complexity, such as crude bio-extracts, intact cells and tissues, fibres, foods, feeds, fertilizers and environmental samples. Here we outline examples to promote an understanding of recent applications of solution-state, solid-state, time-domain NMR and magnetic resonance imaging (MRI) to the complex evaluation of organisms, materials and the environment. We also describe useful databases and informatics tools, as well as machine learning techniques for NMR analysis, demonstrating that NMR data science can be used to evaluate the exposome in both the natural environment and human society towards a sustainable future.

An Improved Stacked Autoencoder for Metabolomic Data Classification

Naru3 (NR) is a traditional Mongolian medicine with high clinical efficacy and low incidence of side effects. Metabolomics is an approach that can facilitate the development of traditional drugs. However, metabolomic data have a high throughput, sparse, high-dimensional, and small sample nature, and their classification is challenging. Although deep learning methods have a wide range of applications, deep learning-based metabolomic studies have not been widely performed. We aimed to develop an improved stacked autoencoder (SAE) for metabolomic data classification. We established an NR-treated rheumatoid arthritis (RA) mouse model and classified the obtained metabolomic data using the Hessian-free SAE (HF-SAE) algorithm. During training, the unlabeled data were used for pretraining, and the labeled data were used for fine-tuning based on the HF algorithm for gradient descent optimization. The hybrid algorithm successfully classified the data. The results were compared with those of the support vector machine (SVM), k-nearest neighbor (KNN), and gradient descent SAE (GD-SAE) algorithms. A five-fold cross-validation was used to complete the classification experiment. In each fine-tuning process, the mean square error (MSE) and misclassification rates of the training and test data were recorded. We successfully established an NR animal model and an improved SAE for metabolomic data classification.

Relaxometric learning: a pattern recognition method for T2 relaxation curves based on machine learning supported by an analytical framework

Nuclear magnetic resonance (NMR)-based relaxometry is widely used in various fields of research because of its advantages such as simple sample preparation, easy handling, and relatively low cost compared with metabolomics approaches. However, there have been no reports on the application of the T₂ relaxation curves in metabolomics studies involving the evaluation of metabolic mixtures, such as geographical origin determination and feature extraction by pattern recognition and data mining. In this study, we describe a data mining method for relaxometric data (i.e., relaxometric learning). This method is based on a machine learning algorithm supported by the analytical framework optimized for the relaxation curve analyses. In the analytical framework, we incorporated a variable optimization approach and bootstrap resampling-based matrixing to enhance the classification performance and balance the sample size between groups, respectively. The relaxometric learning enabled the extraction of features related to the physical properties of fish muscle and the determination of the geographical origin of the fish by improving the classification performance. Our results suggest that relaxometric learning is a powerful and versatile alternative to conventional metabolomics approaches for evaluating fleshiness of chemical mixtures in food and for other biological and chemical research requiring a nondestructive, cost-effective, and time-saving method.

Fish ecotyping based on machine learning and inferred network analysis of chemical and physical properties

Functional diversity rather than species richness is critical for the understanding of ecological patterns and processes. This study aimed to develop novel integrated analytical strategies for the functional characterization of fish diversity based on the quantification, prediction and integration of the chemical and physical features in fish muscles. Machine learning models with an improved random forest algorithm applied on 1867 muscle nuclear magnetic resonance spectra belonging to 249 fish species successfully predicted the mobility patterns of fishes into four categories (migratory, territorial, rockfish, and demersal) with accuracies of 90.3-95.4%. Markov blanket-based feature selection method with an ecological-chemical-physical integrated network based on the Bayesian network inference algorithm highlighted the importance of nitrogen metabolism, which is critical for environmental adaptability of fishes in nutrient-rich environments, in the functional characterization of fish biodiversity. Our study provides valuable information and analytical strategies for fish home-range assessment on the basis of the chemical and physical characterization of fish muscle, which can serve as an ecological indicator for fish ecotyping and human impact monitoring.

Prediction of Liquid Chromatographic Retention Time with Graph Neural Networks to Assist in Small Molecule Identification

The predicted liquid chromatographic retention times (RTs) of small molecules are not accurate enough for wide adoption in structural identification. In this study, we used the graph neural network to predict the retention time (GNN-RT) from structures of small molecules directly without the requirement of molecular descriptors. The predicted accuracy of GNN-RT was compared with random forests (RFs), Bayesian ridge regression, convolutional neural network (CNN), and a deep-learning regression model (DLM) on a METLIN small molecule retention time (SMRT) dataset. GNN-RT achieved the highest predicting accuracy with a mean relative error of 4.9% and a median relative error of 3.2%. Furthermore, the SMRT-trained GNN-RT model can be transferred to the same type of chromatographic systems easily. The predicted RT is valuable for structural identification in complementary to tandem mass spectra and can be used to assist in the identification of compounds. The results indicate that GNN-RT is a promising method to predict the RT for liquid chromatography and improve the accuracy of structural identification for small molecules.

Deep metabolome: Applications of deep learning in metabolomics

In the past few years, deep learning has been successfully applied to various omics data. However, the applications of deep learning in metabolomics are still relatively low compared to others omics. Currently, data pre-processing using convolutional neural network architecture appears to benefit the most from deep learning. Compound/structure identification and quantification using artificial neural network/deep learning performed relatively better than traditional machine learning techniques, whereas only marginally better results are observed in biological interpretations. Before deep learning can be effectively applied to metabolomics, several challenges should be addressed, including metabolome-specific deep learning architectures, dimensionality problems, and model evaluation regimes.

Deep learning meets metabolomics: a methodological perspective

Deep learning (DL), an emerging area of investigation in the fields of machine learning and artificial intelligence, has markedly advanced over the past years. DL techniques are being applied to assist medical professionals and researchers in improving clinical diagnosis, disease prediction and drug discovery. It is expected that DL will help to provide actionable knowledge from a variety of 'big data', including metabolomics data. In this review, we discuss the applicability of DL to metabolomics, while presenting and discussing several examples from recent research. We emphasize the use of DL in tackling bottlenecks in metabolomics data acquisition, processing, metabolite identification, as well as in metabolic phenotyping and biomarker discovery. Finally, we discuss how DL is used in genome-scale metabolic modelling and in interpretation of metabolomics data. The DL-based approaches discussed here may assist computational biologists with the integration, prediction and drawing of statistical inference about biological outcomes, based on metabolomics data.

Large-Scale Evaluation of Major Soluble Macromolecular Components of Fish Muscle from a Conventional 1H-NMR Spectral Database

Conventional proton nuclear magnetic resonance (¹H-NMR) has been widely used for identification and quantification of small molecular components in food. However, identification of major soluble macromolecular components from conventional ¹H-NMR spectra is difficult. This is because the baseline appearance is masked by the dense and high-intensity signals from small molecular components present in the sample mixtures. In this study, we introduced an integrated analytical strategy based on the combination of additional measurement using a diffusion filter, covariation peak separation, and matrix decomposition in a small-scale training dataset. This strategy is aimed to extract signal profiles of soluble macromolecular components from conventional ¹H-NMR spectral data in a large-scale dataset without the requirement of re-measurement. We applied this method to the conventional ¹H-NMR spectra of water-soluble fish muscle extracts and investigated the distribution characteristics of fish diversity and muscle soluble macromolecular components, such as lipids and collagens. We identified a cluster of fish species with low content of lipids and high content of collagens in muscle, which showed great potential for the development of functional foods. Because this mechanical data processing method requires additional measurement of only a small-scale training dataset without special sample pretreatment, it should be immediately applicable to extract macromolecular signals from accumulated conventional ¹H-NMR databases of other complex gelatinous mixtures in foods.

Cell metabolomics to study the cytotoxicity of carbon black nanoparticles on A549 cells using UHPLC-Q/TOF-MS and multivariate data analysis

Carbon black nanoparticles (CBNPs) are core component of fine particulate matter (PM2.5) in the atmosphere. It was reported that the particle in the atmosphere with smaller size and the larger the specific surface area are easier to reach the deep respiratory tract or even the alveoli through the respiratory barrier and cause lung injury. Therefore, it has been believed that ultrafine or nanometer particles with more toxic than those with larger particle sizes. Moreover, it was confirmed that CBNPs could induce inflammation, oxidative stress and changes in cell signaling and gene expression in mammalian cells and organs. However, the cytotoxicity mechanism of them has been uncertain so far. The aim of the present study was to explore the underlying mechanism of cytotoxicity induced by CBNPs on A549 cells. In the current research, the viabilities of A549 cells were detected by Cell Counting Kit-8 (CCK-8) assay. The further metabolomics studies were conducted to detect the cytotoxic effect of CBNPs on A549 cells with an IC50 value of 70 μg/mL for 48 h. Potential differential compounds were identified and quantified using a novel on-line acquisition method based on ultra-liquid chromatography quadrupole time-of-flight mass spectrometry(UHPLC-Q-TOF/MS). The cytotoxicity mechanism of CBNPs on A549 cells was evaluated by multivariate data analysis and statistics. As a result, a total of 32 differential compounds were identified between CBNPs exposure and control groups. In addition, pathway analysis showed the metabolic changes were involved in the tricarboxylic acid (TCA) cycle, alanine, aspartate and glutamate metabolism, histidine metabolism and so on. It is also suggested that CBNPs may induce cytotoxicity by affecting the normal process of energy metabolism and disturbing several vital signaling pathways and finally induce cell apoptosis.

The application of artificial neural networks in metabolomics: a historical perspective

BACKGROUND

Metabolomics data, with its complex covariance structure, is typically modelled by projection-based machine learning (ML) methods such as partial least squares (PLS) regression, which project data into a latent structure. Biological data are often non-linear, so it is reasonable to hypothesize that metabolomics data may also have a non-linear latent structure, which in turn would be best modelled using non-linear equations. A non-linear ML method with a similar projection equation structure to PLS is artificial neural networks (ANNs). While ANNs were first applied to metabolic profiling data in the 1990s, the lack of community acceptance combined with limitations in computational capacity and the lack of volume of data for robust non-linear model optimisation inhibited their widespread use. Due to recent advances in computational power, modelling improvements, community acceptance, and the more demanding needs for data science, ANNs have made a recent resurgence in interest across research communities, including a small yet growing usage in metabolomics. As metabolomics experiments become more complex and start to be integrated with other omics data, there is potential for ANNs to become a viable alternative to linear projection methods.

AIM OF REVIEW

We aim to first describe ANNs and their structural equivalence to linear projection-based methods, including PLS regression. We then review the historical, current, and future uses of ANNs in the field of metabolomics.

KEY SCIENTIFIC CONCEPT OF REVIEW

Is metabolomics ready for the return of artificial neural networks?

Convolutional neural network for hyperspectral data analysis and effective wavelengths selection

Fusion of spectral and spatial information has been proved to be an effective approach to improve model performance in near-infrared hyperspectral data analysis. Regardless, most of the existing spectral-spatial classification methods require fairly complex pipelines and exact selection of parameters, which mainly depend on the investigator's experience and the object under test. Convolutional neural network (CNN) is a powerful tool for representing complicated data and usually works with few "hand-engineering", making it an appropriate candidate for developing a general and automatic approach. In this paper, a two-branch convolutional neural network (2B-CNN) was developed for spectral-spatial classification and effective wavelengths (EWs) selection. The proposed network was evaluated by three classification data sets, including herbal medicine, coffee bean and strawberry. The results showed that the 2B-CNN obtained the best classification accuracies (96.72% in average) when compared with support vector machine (92.60% in average), one dimensional CNN (92.58% in average), and grey level co-occurrence matrix based support vector machine (93.83% in average). Furthermore, the learned weights of the two-dimensional branch in 2B-CNN were adopted as the indicator of EWs and compared with the successive projections algorithm. The 2B-CNN models built with wavelengths selected by the weight indicator achieved the best accuracies (96.02% in average) among all the examined EWs models. Different from the conventional EWs selection method, the proposed algorithm works without any additional retraining and has the ability to comprehensively consider the discriminative power in spectral domain and spatial domain.

Metabolomic Strategies in Biomarker Research-New Approach for Indirect Identification of Drug Consumption and Sample Manipulation in Clinical and Forensic Toxicology?

Drug of abuse (DOA) consumption is a growing problem worldwide, particularly with increasing numbers of new psychoactive substances (NPS) entering the drug market. Generally, little information on their adverse effects and toxicity are available. The direct detection and identification of NPS is an analytical challenge due to their ephemerality on the drug scene. An approach that does not directly focus on the structural detection of an analyte or its metabolites, would be beneficial for this complex analytical scenario and the development of alternative screening methods could help to provide fast response on suspected NPS consumption. A metabolomics approach might represent such an alternative strategy for the identification of biomarkers for different questions in DOA testing. Metabolomics is the monitoring of changes in small (endogenous) molecules (<1,000 Da) in response to a certain stimulus, e.g., DOA consumption. For this review, a literature search targeting "metabolomics" and different DOAs or NPS was conducted. Thereby, different applications of metabolomic strategies in biomarker research for DOA identification were identified: (a) as an additional tool for metabolism studies bearing the major advantage that particularly a priori unknown or unexpected metabolites can be identified; and (b) for identification of endogenous biomarker or metabolite patterns, e.g., for synthetic cannabinoids or also to indirectly detect urine manipulation attempts by chemical adulteration or replacement with artificial urine samples. The majority of the currently available literature in that field, however, deals with metabolomic studies for DOAs to better assess their acute or chronic effects or to find biomarkers for drug addiction and tolerance. Certain changes in endogenous compounds are detected for all studied DOAs, but often similar compounds/pathways are influenced. When evaluating these studies with regard to possible biomarkers for drug consumption, the observed changes appear, albeit statistically significant, too small to reliably work as biomarker for drug consumption. Further, different drugs were shown to affect the same pathways. In conclusion, metabolomic approaches possess potential for detection of biomarkers indicating drug consumption. More studies, including more sensitive targeted analyses, multi-variant statistical models or deep-learning approaches are needed to fully explore the potential of omics science in DOA testing.

Tools and resources for metabolomics research community: A 2017-2018 update

The scale at which MS- and NMR-based platforms generate metabolomics datasets for both research, core, and clinical facilities to address challenges in the various sciences-ranging from biomedical to agricultural-is underappreciated. Thus, metabolomics efforts spanning microbe, environment, plant, animal, and human systems have led to continual and concomitant growth of in silico resources for analysis and interpretation of these datasets. These software tools, resources, and databases drive the field forward to help keep pace with the amount of data being generated and the sophisticated and diverse analytical platforms that are being used to generate these metabolomics datasets. To address challenges in data preprocessing, metabolite annotation, statistical interrogation, visualization, interpretation, and integration, the metabolomics and informatics research community comes up with hundreds of tools every year. The purpose of the present review is to provide a brief and useful summary of more than 95 metabolomics tools, software, and databases that were either developed or significantly improved during 2017-2018. We hope to see this review help readers, developers, and researchers to obtain informed access to these thorough lists of resources for further improvisation, implementation, and application in due course of time.