Using Expert Driven Machine Learning to Enhance Dynamic Metabolomics Data Analysis

Machine learning for the advancement of genome-scale metabolic modeling

Constraint-based modeling (CBM) has evolved as the core systems biology tool to map the interrelations between genotype, phenotype, and external environment. The recent advancement of high-throughput experimental approaches and multi-omics strategies has generated a plethora of new and precise information from wide-ranging biological domains. On the other hand, the continuously growing field of machine learning (ML) and its specialized branch of deep learning (DL) provide essential computational architectures for decoding complex and heterogeneous biological data. In recent years, both multi-omics and ML have assisted in the escalation of CBM. Condition-specific omics data, such as transcriptomics and proteomics, helped contextualize the model prediction while analyzing a particular phenotypic signature. At the same time, the advanced ML tools have eased the model reconstruction and analysis to increase the accuracy and prediction power. However, the development of these multi-disciplinary methodological frameworks mainly occurs independently, which limits the concatenation of biological knowledge from different domains. Hence, we have reviewed the potential of integrating multi-disciplinary tools and strategies from various fields, such as synthetic biology, CBM, omics, and ML, to explore the biochemical phenomenon beyond the conventional biological dogma. How the integrative knowledge of these intersected domains has improved bioengineering and biomedical applications has also been highlighted. We categorically explained the conventional genome-scale metabolic model (GEM) reconstruction tools and their improvement strategies through ML paradigms. Further, the crucial role of ML and DL in omics data restructuring for GEM development has also been briefly discussed. Finally, the case-study-based assessment of the state-of-the-art method for improving biomedical and metabolic engineering strategies has been elaborated. Therefore, this review demonstrates how integrating experimental and in silico strategies can help map the ever-expanding knowledge of biological systems driven by condition-specific cellular information. This multiview approach will elevate the application of ML-based CBM in the biomedical and bioengineering fields for the betterment of society and the environment.

The HuMet Repository: Watching human metabolism at work

The human metabolism constantly responds to stimuli such as food intake, fasting, exercise, and stress, triggering adaptive biochemical processes across multiple metabolic pathways. To understand the role of these processes and disruptions thereof in health and disease, detailed documentation of healthy metabolic responses is needed but still scarce on a time-resolved metabolome-wide level. Here, we present the HuMet Repository, a web-based resource for exploring dynamic metabolic responses to six physiological challenges (exercise, 36 h fasting, oral glucose and lipid loads, mixed meal, cold stress) in healthy subjects. For building this resource, we integrated existing and newly derived metabolomics data measured in blood, urine, and breath samples of 15 young healthy men at up to 56 time points during the six highly standardized challenge tests conducted over four days. The data comprise 1.1 million data points acquired on multiple platforms with temporal profiles of 2,656 metabolites from a broad range of biochemical pathways. By embedding the dataset into an interactive web application, we enable users to easily access, search, filter, analyze, and visualize the time-resolved metabolomic readouts and derived results. Users can put metabolites into their larger context by identifying metabolites with similar trajectories or by visualizing metabolites within holistic metabolic networks to pinpoint pathways of interest. In three showcases, we outline the value of the repository for gaining biological insights and generating hypotheses by analyzing the wash-out of dietary markers, the complementarity of metabolomics platforms in dynamic versus cross-sectional data, and similarities and differences in systemic metabolic responses across challenges. With its comprehensive collection of time-resolved metabolomics data, the HuMet Repository, freely accessible at https://humet.org/, is a reference for normal, healthy responses to metabolic challenges in young males. It will enable researchers with and without computational expertise, to flexibly query the data for their own research into the dynamics of human metabolism.

In Vitro Biotransformation and Anti-Inflammatory Activity of Constituents and Metabolites of Filipendula ulmaria

(1) Background: Filipendula ulmaria (L.) Maxim. (Rosaceae) (meadowsweet) is widely used in phytotherapy against inflammatory diseases. However, its active constituents are not exactly known. Moreover, it contains many constituents, such as flavonoid glycosides, which are not absorbed, but metabolized in the colon by gut microbiota, producing potentially active metabolites that can be absorbed. The aim of this study was to characterize the active constituents or metabolites. (2) Methods: A F. ulmaria extract was processed in an in vitro gastrointestinal biotransformation model, and the metabolites were characterized using UHPLC-ESI-QTOF-MS analysis. In vitro anti-inflammatory activity was evaluated by testing the inhibition of NF-κB activation, COX-1 and COX-2 enzyme inhibition. (3) Results: The simulation of gastrointestinal biotransformation showed a decrease in the relative abundance of glycosylated flavonoids such as rutin, spiraeoside and isoquercitrin in the colon compartment, and an increase in aglycons such as quercetin, apigenin, naringenin and kaempferol. The genuine as well as the metabolized extract showed a better inhibition of the COX-1 enzyme as compared to COX-2. A mix of aglycons present after biotransformation showed a significant inhibition of COX-1. (4) Conclusions: The anti-inflammatory activity of F. ulmaria may be explained by an additive or synergistic effect of genuine constituents and metabolites.

Antiplasmodial activity of constituents and their metabolites after in vitro gastrointestinal biotransformation of a Nauclea pobeguinii extract

Nauclea pobeguinii is traditionally used for treatment of malaria. Previous studies on the plant extract and strictosamide, the putative active constituent, showed a profound in vivo activity of the extract but no in vitro activity of strictosamide. This might indicate that one or more compounds present in the extract, most likely alkaloids, act as prodrugs undergoing biotransformation after oral administration resulting in the active compounds. The phytochemical composition of a N. pobeguinii extract was characterized using UHPLC-UV-HRMS (Ultrahigh-Performance Liquid Chromatography-Ultraviolet-High Resolution Mass Spectrometry) data. An in vitro gastrointestinal model was used to simulate biotransformation of the extract allowing monitoring of the relative abundances of individual constituents over time on one hand, while antiplasmodial activity and cytotoxicity of the biotransformed extract could be evaluated on the other hand. A diversity of compounds was (tentatively) identified in the extract, mainly saponins and alkaloids, including 32 compounds that have not been reported before in N. pobeguinii. The automated data analysis workflow used for unbiased screening for metabolites showed that glycosylated compounds decreased in intensity over time. Alkaloids containing no sugar moieties, including angustine-type alkaloids, showed no gastrointestinal biotransformation. In vitro gastrointestinal biotransformation of strictosamide did not result in a major metabolite. Moreover, multivariate data analysis using Orthogonal Partial Least Square-Discriminant Analysis (OPLS-DA) showed no in vitro activity of strictosamide or its metabolites suggesting that other compounds or metabolites present in the extract are responsible for the antiplasmodial effect of the N. pobeguinii extract. The OPLS-DA proposes alkaloids with a β-carboline moiety as active principles, suggesting that antiplasmodial activity of N. pobeguinii derives from an additive or synergistic effect of multiple minor alkaloids and their metabolites present in the bark extract of N. pobeguinii.

Simulated Gastrointestinal Biotransformation of Chlorogenic Acid, Flavonoids, Flavonolignans and Triterpenoid Saponins in Cecropia obtusifolia Leaf Extract

It is well known that biotransformation processes in the human body are crucial to form potentially bioactive metabolites from particular classes of natural products. However, little research has been conducted concerning the bioavailability of polyphenols, especially in the colon. The gastrointestinal stability and colonic biotransformation of the crude extract of the leaves of Cecropia obtusifolia, rich in flavone C-glycosides, was investigated under in vitro conditions, and the processing and interpretation of results were facilitated by using an automated machine learning model. This investigation revealed that flavone C-glycosides and flavonolignans from C. obtusifolia were stable throughout their passage in the simulated gastrointestinal tract including the colon phase. On the other hand, the colon bacteria extensively metabolized chlorogenic acid, flavonol, and triterpenoid O-glycosides. This investigation revealed that the colonic microbiota has an important role in the biotransformation of some chemical constituents of this extract.

An exploratory approach for an oriented development of an untargeted hydrophilic interaction liquid chromatography-mass spectrometry platform for polar metabolites in biological matrices

The analysis of polar metabolites based on liquid chromatography-mass spectrometry (LC-MS) methods should take into consideration the complexity of interactions in LC columns to be able to cover a broad range of metabolites of key biological pathways. Therefore, in this study, different chromatographic columns were tested for polar metabolites including reversed-phase and hydrophilic interaction liquid chromatography (HILIC) columns. Based on a column screening, two new generations of zwitterionic HILIC columns were selected for further evaluation. A tree-based method optimization was applied to investigate the chromatographic factors affecting the retention mechanisms of polar metabolites with zwitterionic stationary phases. The results were evaluated based on a scoring system which was applied for more than 80 polar metabolites with a high coverage of key human metabolic pathways. The final optimized methods showed high complementarity to analyze a wide range of metabolic classes including amino acids, small peptides, sugars, amino sugars, phosphorylated sugars, organic acids, nucleobases, nucleosides, nucleotides and acylcarnitines. Optimized methods were applied to analyze different biological matrices, including human urine, plasma and liver cell extracts using an untargeted approach. The number of high-quality features (< 30% median relative standard deviation) ranged from 3,755 for urine to 5,402 for the intracellular metabolome of liver cells, showing the potential of the methods for untargeted purposes.

A comparative study on the in vitro biotransformation of medicagenic acid using human liver microsomes and S9 fractions

Many natural products are prodrugs which are biotransformed and activated after oral administration. The investigation of gastrointestinal and hepatic biotransformation can be facilitated by in vitro screening methods. This study compares two widely used in vitro models for hepatic biotransformation: 1) human S9 fractions and 2) human liver microsomes and cytosolic fractions in a two-step sequence, with the purpose of identifying differences in the biotransformation of medicagenic acid, the putative precursor of active metabolites, responsible for the medicinal effects of the herb Herniaria hirsuta. The combination of liquid chromatography coupled to high-resolution mass spectrometry with subsequent suspect and non-target data analysis allowed the identification of thirteen biotransformation products, four of which are reported here for the first time. Eight biotransformation products resulting from oxidative Phase I reactions were identified. Phase II conjugation reactions resulted in the formation of three glucuronidated and two sulfated biotransformation products. No major differences could be observed between incubations with human liver S9 or when utilizing human microsomal and cytosolic fractions. Apart from two metabolites, both methods rendered the same qualitative metabolic profile, with minor quantitative differences. As a result, both protocols applied in this study can be used to study in vitro human liver biotransformation reactions.

Discovery of different metabotypes in overconditioned dairy cows by means of machine learning

Using data from targeted metabolomics in serum in combination with machine learning (ML) approaches, we aimed at (1) identifying divergent metabotypes in overconditioned cows and at (2) exploring how metabotypes are associated with lactation performance, blood metabolites, and hormones. In a previously established animal model, 38 pregnant multiparous Holstein cows were assigned to 2 groups that were fed differently to reach either high (HBCS) or normal (NBCS) body condition score (BCS) and backfat thickness (BFT) until dryoff at -49 d before calving [NBCS: BCS < 3.5 (3.02 ± 0.24) and BFT < 1.2 cm (0.92 ± 0.21), mean ± SD; HBCS: BCS > 3.75 (3.82 ± 0.33) and BFT > 1.4 cm (2.36 ± 0.35)]. Cows were then fed the same diets during the dry period and the subsequent lactation, and maintained the differences in BFT and BCS throughout the study. Blood samples were collected weekly from 7 wk antepartum (ap) to 12 wk postpartum (pp) to assess serum concentrations of metabolites (by targeted metabolomics and by classical analyses) and metabolic hormones. Metabolic clustering by applying 4 supervised ML-based classifiers [sequential minimal optimization (SMO), random forest (RF), alternating decision tree (ADTree), and naïve Bayes-updatable (NB)] on the changes (d 21 pp minus d 49 ap) in concentrations of 170 serum metabolites resulted in 4 distinct metabolic clusters: HBCS predicted HBCS (HBCS-PH, n = 13), HBCS predicted NBCS (HBCS-PN, n = 6), NBCS predicted NBCS (NBCS-PN, n = 15), and NBCS predicted HBCS (NBCS-PH, n = 4). The accuracies of SMO, RF, ADTree, and NB classifiers were >70%. Because the number of NBCS-PH cows was low, we did not consider this group for further comparisons. Dry matter intake (kg/d and percentage of body weight) and energy intake were greater in HBCS-PN than in HBCS-PH in early lactation, and HBCS-PN also reached a positive energy balance earlier than did HBCS-PH. Milk yield was not different between groups, but milk protein percentage was greater in HBCS-PN than in HBCS-PH cows. The circulating concentrations of fatty acids (FA) increased during early lactation in both groups, but HBCS-PN cows had lower concentrations of β-hydroxybutyrate, indicating lower ketogenesis compared with HBCS-PH cows. The concentrations of insulin, insulin-like growth factor 1, leptin, adiponectin, haptoglobin, glucose, and revised quantitative insulin sensitivity check index did not differ between the groups, whereas serum concentrations of glycerophospholipids were lower before calving in HBCS-PH than in HBCS-PN cows. Glycine was the only amino acid that had higher concentration after calving in HBCS-PH than in HBCS-PN cows. The circulating concentrations of some short- (C2, C3, and C4) and long-chain (C12, C16:0, C18:0, and C18:1) acylcarnitines on d 21 pp were greater in HBCS-PH than in HBCS-PN cows, indicating incomplete FA oxidation. In conclusion, the use of ML approaches involving data from targeted metabolomics in serum is a promising method for differentiating divergent metabotypes from apparently similar BCS phenotypes. Further investigations, using larger numbers of cows and farms, are warranted for confirmation of this finding.

Compound Characterization and Metabolic Profile Elucidation after In Vitro Gastrointestinal and Hepatic Biotransformation of an Herniaria hirsuta Extract Using Unbiased Dynamic Metabolomic Data Analysis

Herniaria hirsuta L. (Caryophyllaceae) is used for treatment of urinary stones and as a diuretic. Little is known about the active compounds and the mechanism of action. The phytochemical composition of H. hirsuta was comprehensively characterized using UHPLC-UV-HRMS (Ultrahigh-Performance Liquid Chromatography-Ultraviolet-High Resolution Mass Spectrometry) data. An in vitro gastrointestinal model was used to simulate biotransformation, which allowed the monitoring of the relative abundances of individual compounds over time. To analyze the longitudinal multiclass LC-MS data, XCMS, a platform that enables online metabolomics data processing and interpretation, and EDGE, a statistical method for time series data, were used to extract significant differential profiles from the raw data. An interactive Shiny app in R was used to rate the quality of the resulting features. These ratings were used to train a random forest model. The most abundant aglycone after gastrointestinal biotransformation was subjected to hepatic biotransformation using human S9 fractions. A diversity of compounds was detected, mainly saponins and flavonoids. Besides the known saponins, 15 new saponins were tentatively identified as glycosides of medicagenic acid, acetylated medicagenic acid and zanhic acid. It is suggested that metabolites of phytochemicals present in H. hirsuta, most likely saponins, are responsible for the pharmaceutical effects. It was observed that the relative abundance of saponin aglycones increased, indicating loss of sugar moieties during colonic biotransformation, with medicagenic acid as the most abundant aglycone. Hepatic biotransformation of this aglycone resulted in different metabolites formed by phase I and II reactions.

Toward a Standardized Strategy of Clinical Metabolomics for the Advancement of Precision Medicine

Despite the tremendous success, pitfalls have been observed in every step of a clinical metabolomics workflow, which impedes the internal validity of the study. Furthermore, the demand for logistics, instrumentations, and computational resources for metabolic phenotyping studies has far exceeded our expectations. In this conceptual review, we will cover inclusive barriers of a metabolomics-based clinical study and suggest potential solutions in the hope of enhancing study robustness, usability, and transferability. The importance of quality assurance and quality control procedures is discussed, followed by a practical rule containing five phases, including two additional "pre-pre-" and "post-post-" analytical steps. Besides, we will elucidate the potential involvement of machine learning and demonstrate that the need for automated data mining algorithms to improve the quality of future research is undeniable. Consequently, we propose a comprehensive metabolomics framework, along with an appropriate checklist refined from current guidelines and our previously published assessment, in the attempt to accurately translate achievements in metabolomics into clinical and epidemiological research. Furthermore, the integration of multifaceted multi-omics approaches with metabolomics as the pillar member is in urgent need. When combining with other social or nutritional factors, we can gather complete omics profiles for a particular disease. Our discussion reflects the current obstacles and potential solutions toward the progressing trend of utilizing metabolomics in clinical research to create the next-generation healthcare system.