Random Forests machine learning applied to gas chromatography – Mass spectrometry derived average mass spectrum data sets for classification and characterisation of essential oils

Revealing quality chemicals of Tetrastigma hemsleyanum roots in different geographical origins using untargeted metabolomics and random-forest based spectrum-effect analysis

Tetrastigma hemsleyanum root is a popular functional food in China, and the price varies based on the origin of the product. The link between the origin, metabolic profile, and bioactivity of T. hemsleyanum must be investigated. This study compares the metabolic profiles of 254 samples collected from eight different areas with 49 potential key chemical markers using plant metabolomics. The metabolic pathways of the five critical flavonoid metabolites were annotated and enriched using the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway. Moreover, a random forest model aiding the spectrum-effect relationship analysis was developed for the first time indicating catechin and darendoside B as potential quality markers of antioxidant activity. The findings of this study provide a comprehensive understanding of the chemical composition and bioactive compounds of T. hemsleyanum as well as valuable information on the evaluation of the quality of various samples and products in the market.

Voltammetric techniques for low-cost on-site routine analysis of thymol in the medicinal plant Ocimum gratissimum

The composition of essential oils varies according to culture conditions and climate, which induces a need for simple and inexpensive characterization methods close to the place of extraction. This appears particularly important for developing countries. Herein, we develop an analytical strategy to determine the thymol content in Ocimum Gratissimum, a medicinal plant from Benin. The protocol is based on electrochemical techniques (cyclic and square wave voltammetry) implemented with a low cost potentiostat. Thymol is a phenol derivative and was directly oxidized at the electrode surface. We had to resort to submillimolar concentrations (25-300 μM) in order to minimize production of phenol oligomers that passivate the electrode. We worked first on two essential oils and realized that in one of them the thymol concentration was below our detection method. These results were confirmed by gas chromatography - mass spectrometry. Furthermore, we optimized the detection protocol to analyze an infusion made directly from the leaves of the plant. Finally, we studied whether the cost of the electrochemical cell may also be minimized by using pencil lead as working and counter electrodes.

Retention Time Trajectory Matching for Peak Identification in Chromatographic Analysis

Retention time drift caused by fluctuations in physical factors such as temperature ramping rate and carrier gas flow rate is ubiquitous in chromatographic measurements. Proper peak matching and identification across different chromatograms is critical prior to any subsequent analysis but is challenging without using mass spectrometry. The purpose of this work was to describe and validate a peak matching and identification method called retention time trajectory (RTT) matching that can be used in targeted analyses free of mass spectrometry. This method uses chromatographic retention times as the only input and identifies peaks associated with any subset of a predefined set of target compounds. An RTT is a two-dimensional (2D) curve formed uniquely by the retention times of the chromatographic peaks. The RTTs obtained from the chromatogram of a sample under test and those pre-installed in a library are matched and statistically compared. The best matched pair implies identification. Unlike most existing peak-alignment methods, no mathematical warping or transformation is involved. Based on the experimentally characterized RTT, an RTT hybridization method was also developed to rapidly generate more RTTs and expand the library without performing actual time-consuming chromatographic measurements, which enables successful peak matching even for chromatograms with severe retention time drifts. Additionally, 3.15 × 10⁵ tests using experimentally obtained gas chromatograms and 2 × 10¹² tests using two publicly available fruit metabolomics datasets validated the proposed method, demonstrating real-time peak/interferent identification.

Fully Automated Unconstrained Analysis of High-Resolution Mass Spectrometry Data with Machine Learning

Mass spectrometry (MS) is a convenient, highly sensitive, and reliable method for the analysis of complex mixtures, which is vital for materials science, life sciences fields such as metabolomics and proteomics, and mechanistic research in chemistry. Although it is one of the most powerful methods for individual compound detection, complete signal assignment in complex mixtures is still a great challenge. The unconstrained formula-generating algorithm, covering the entire spectra and revealing components, is a "dream tool" for researchers. We present the framework for efficient MS data interpretation, describing a novel approach for detailed analysis based on deisotoping performed by gradient-boosted decision trees and a neural network that generates molecular formulas from the fine isotopic structure, approaching the long-standing inverse spectral problem. The methods were successfully tested on three examples: fragment ion analysis in protein sequencing for proteomics, analysis of the natural samples for life sciences, and study of the cross-coupling catalytic system for chemistry.

Unveiling Chemical Cues of Insect-Tree and Insect-Insect Interactions for the Eucalyptus Weevil and Its Egg Parasitoid by Multidimensional Gas Chromatographic Methods

Multidimensional gas chromatography is, presently, an established and powerful analytical tool, due to higher resolving power than the classical 1D chromatographic approaches. Applied to multiple areas, it allows to isolate, detect and identify a larger number of compounds present in complex matrices, even in trace amounts. Research was conducted to determine which compounds, emitted by host plants of the eucalyptus weevil, Gonipterus platensis, might mediate host selection behavior. The identification of a pheromone blend of G. platensis is presented, revealing to be more attractive to weevils of both sexes, than the individual compounds. The volatile organic compounds (VOCs) were collected by headspace solid phase microextraction (HS-SPME), MonoTrapTM disks, and simultaneous distillation-extraction (SDE). Combining one dimensional (1D) and two-dimensional (2D) chromatographic systems—comprehensive and heart-cut two-dimensional gas chromatography (GC×GC and H/C-MD-GC, respectively) with mass spectrometry (MS) and electroantennographic (EAD) detection, enabled the selection and identification of pertinent semiochemicals which were detected by the insect antennal olfactory system. The behavioral effect of a selected blend of compounds was assessed in a two-arm olfactometer with ten parallel walking chambers, coupled to video tracking and data analysis software. An active blend, composed by cis and trans-verbenol, verbenene, myrtenol and trans-pinocarveol was achieved.

Deep Learning-Assisted Peak Curation for Large-Scale LC-MS Metabolomics

Available automated methods for peak detection in untargeted metabolomics suffer from poor precision. We present NeatMS, which uses machine learning based on a convoluted neural network to reduce the number and fraction of false peaks. NeatMS comes with a pre-trained model representing expert knowledge in the differentiation of true chemical signal from noise. Furthermore, it provides all necessary functions to easily train new models or improve existing ones by transfer learning. Thus, the tool improves peak curation and contributes to the robust and scalable analysis of large-scale experiments. We show how to integrate it into different liquid chromatography–mass spectrometry (LC-MS) analysis workflows, quantify its performance, and compare it to various other approaches. NeatMS software is available as open source on github under permissive MIT license and is also provided as easy-to-install PyPi and Bioconda packages.

Using random forest to detect multiple inherited metabolic diseases simultaneously based on GC-MS urinary metabolomics

Inborn errors of metabolism, also known as inherited metabolic diseases (IMDs), are related to genetic mutations and cause corresponding biochemical metabolic disorder of newborns and even sudden infant death. Timely detection and diagnosis of IMDs are of great significance for improving survival of newborns. Here we propose a strategy for simultaneously detecting six types of IMDs via combining GC-MS technique with the random forest algorithm (RF). Clinical urine samples from IMD and healthy patients are analyzed using GC-MS for acquiring metabolomics data. Then, the RF model is established as a multi-classification tool for the GC-MS data. Compared with the models built by artificial neural network and support vector machine, the results demonstrated the RF model has superior performance of high specificity, sensitivity, precision, accuracy, and matthews correlation coefficients on identifying all six types of IMDs and normal samples. The proposed strategy can afford a useful method for reliable and effective identification of multiple IMDs in clinical diagnosis.

Machine Learning Data Augmentation as a Tool to Enhance Quantitative Composition-Activity Relationships of Complex Mixtures. A New Application to Dissect the Role of Main Chemical Components in Bioactive Essential Oils

Scientific investigation on essential oils composition and the related biological profile are continuously growing. Nevertheless, only a few studies have been performed on the relationships between chemical composition and biological data. Herein, the investigation of 61 assayed essential oils is reported focusing on their inhibition activity against Microsporum spp. including development of machine learning models with the aim of highlining the possible chemical components mainly related to the inhibitory potency. The application of machine learning and deep learning techniques for predictive and descriptive purposes have been applied successfully to many fields. Quantitative composition-activity relationships machine learning-based models were developed for the 61 essential oils tested as Microsporum spp. growth modulators. The models were built with in-house python scripts implementing data augmentation with the purpose of having a smoother flow between essential oils' chemical compositions and biological data. High statistical coefficient values (Accuracy, Matthews correlation coefficient and F1 score) were obtained and model inspection permitted to detect possible specific roles related to some components of essential oils' constituents. Robust machine learning models are far more useful tools to reveal data augmentation in comparison with raw data derived models. To the best of the authors knowledge this is the first report using data augmentation to highlight the role of complex mixture components, in particular a first application of these data will be for the development of ingredients in the dermo-cosmetic field investigating microbial species considering the urge for the use of natural preserving and acting antimicrobial agents.

Development of variance rank initiated-unsupervised sample indexing for gas chromatography-mass spectrometry analysis

Traditional non-targeted chemometric workflows for gas chromatography-mass spectrometry (GC-MS) data rely on using supervised methods, which requires a priori knowledge of sample class membership. Herein, we propose a simple, unsupervised chemometric workflow known as variance rank initiated-unsupervised sample indexing (VRI-USI). VRI-USI discovers analyte peaks exhibiting high relative variance across all samples, followed by k-means clustering on the individual peaks. Based upon how the samples cluster for a given peak, a sample index assignment is provided. Using a probabilistic argument, if the same sample index assignment appears for several discovered peaks, then this outcome strongly suggests that the samples are properly classified by that particular sample index assignment. Thus, relevant chemical differences between the samples have been discovered in an unsupervised fashion. The VRI-USI workflow is demonstrated on three, increasingly difficult datasets: simulations, yeast metabolomics, and human cancer metabolomics. For simulated GC-MS datasets, VRI-USI discovered 85-90% of analytes modeled to vary between sample classes. Nineteen out of 53 peaks in the peak table developed for the yeast metabolome dataset had the same sample index assignments, indicating that those indices are most likely due to class-distinguishing chemical differences. A t-test revealed that 22 out of 53 peaks were statistically significant (p < 0.05) when using those sample index assignments. Likewise, for the human cancer metabolomics study, VRI-USI discovered 25 analytes that were statistically different (p < 0.05) using the sample index assignments determined to highlight meaningful sample-based differences. For all datasets, the sample index assignments that were deduced from VRI-USI were the correct class-based difference when using prior knowledge. VRI-USI holds promise as an exploratory data analysis workflow for studies in which analysts do not readily have a priori class information or want to uncover the underlying nature of their dataset.

Smartphone-based handheld Raman spectrometer and machine learning for essential oil quality evaluation

We present a method, utilising a smartphone-based miniaturized Raman spectrometer and machine learning for the fast identification and discrimination of adulterated essential oils (EOs). Firstly, the approach was evaluated for discrimination of pure EOs from those adulterated with solvent, namely benzyl alcohol. In the case of ylang-ylang EO, three different types of adulteration were examined, adulteration with solvent, cheaper vegetable oil and a lower price EO. Random Forest and partial least square discrimination analysis (PLS-DA) showed excellent performance in discriminating pure from adulterated EOs, whilst the same time identifying the type of adulteration. Also, utilising partial least squares regression analysis (PLS) all adulterants, namely benzyl alcohol, vegetable oil and lower price EO, were quantified based on spectra recorded using the smartphone Raman spectrometer, with relative error of prediction (REP) being between 2.41-7.59%.

Leveraging 16S rRNA Microbiome Sequencing Data to Identify Bacterial Signatures for Irritable Bowel Syndrome

Irritable bowel syndrome (IBS) is a chronic gastrointestinal disorder characterized by abdominal pain or discomfort. Previous studies have illustrated that the gut microbiota might play a critical role in IBS, but the conclusions of these studies, based on various methods, were almost impossible to compare, and reproducible microorganism signatures were still in question. To cope with this problem, previously published 16S rRNA gene sequencing data from 439 fecal samples, including 253 IBS samples and 186 control samples, were collected and processed with a uniform bioinformatic pipeline. Although we found no significant differences in community structures between IBS and healthy controls at the amplicon sequence variants (ASV) level, machine learning (ML) approaches enabled us to discriminate IBS from healthy controls at genus level. Linear discriminant analysis effect size (LEfSe) analysis was subsequently used to seek out 97 biomarkers across all studies. Then, we quantified the standardized mean difference (SMDs) for all significant genera identified by LEfSe and ML approaches. Pooled results showed that the SMDs of nine genera had statistical significance, in which the abundance of Lachnoclostridium, Dorea, Erysipelatoclostridium, Prevotella 9, and Clostridium sensu stricto 1 in IBS were higher, while the dominant abundance genera of healthy controls were Ruminococcaceae UCG-005, Holdemanella, Coprococcus 2, and Eubacterium coprostanoligenes group. In summary, based on six published studies, this study identified nine new microbiome biomarkers of IBS, which might be a basis for understanding the key gut microbes associated with IBS, and could be used as potential targets for microbiome-based diagnostics and therapeutics.

Polymeric Carriers Designed for Encapsulation of Essential Oils with Biological Activity

The article reviews the possibilities of encapsulating essential oils EOs, due to their multiple benefits, controlled release, and in order to protect them from environmental conditions. Thus, we present the natural polymers and the synthetic macromolecular chains that are commonly used as networks for embedding EOs, owing to their biodegradability and biocompatibility, interdependent encapsulation methods, and potential applicability of bioactive blend structures. The possibilities of using artificial intelligence to evaluate the bioactivity of EOs—in direct correlation with their chemical constitutions and structures, in order to avoid complex laboratory analyses, to save money and time, and to enhance the final consistency of the products—are also presented.

Identification of metabolic kinetic patterns in different brain regions using metabolomics methods coupled with various discriminant approaches

Metabolomics is widely used as a powerful technique for identifying metabolic patterns and functions of organs and biological systems. Normally, there are multiple groups/targets involved in data processed by discriminant analysis. This is more common in cerebral studies, as there are always several brain regions involved in neuronal studies or brain metabolic dysfunctions. Furthermore, neuronal activity is highly correlated with cerebral energy metabolism, such as oxidation of glucose, especially for glutamatergic (excitatory) and GABAergic (inhibitory) neuronal activities. Thus, regional cerebral energy metabolism recognition is essential for understanding brain functions. In the current study, ten different brain regions were considered for discrimination analysis. The metabolic kinetics were investigated with ¹³C enrichments in metabolic products of glucose and measured using the nuclear magnetic spectroscopic method. Multiple discriminative methods were used to construct classification models in order to screen out the best method. After comparing all the applied discriminatory analysis methods, the boost-decision tree method was found to be the best method for classification and every cerebral region exhibited its own metabolic pattern. Finally, the differences in metabolic kinetics among these brain regions were analyzed. We, therefore, concluded that the current technology could also be utilized in other multi-class metabolomics studies and special metabolic kinetic patterns could provide useful information for brain function studies.

Screening of potential oestrogen receptor α agonists in pesticides via in silico, in vitro and in vivo methods

In modern agricultural management, the use of pesticides is indispensable. Due to their massive use worldwide, pesticides represent a latent risk to both humans and the environment. In the present study, 1056 frequently used pesticides were screened for oestrogen receptor (ER) agonistic activity by using in silico methods. We found that 72 and 47 pesticides potentially have ER agonistic activity by the machine learning methods random forest (RF) and deep neural network (DNN), respectively. Among endocrine-disrupting chemicals (EDCs), 14 have been reported as EDCs or ER agonists by previous studies. We selected 3 reported and 7 previously unreported pesticides from 76 potential ER agonists to further assess ERα agonistic activity. All 10 selected pesticides exhibited ERα agonistic activity in human cells or zebrafish. In the dual-luciferase reporter gene assays, six pesticides exhibited ERα agonistic activity. Additionally, nine pesticides could induce mRNA expression of the pS2 and NRF1 genes in MCF-7 cells, and seven pesticides could induce mRNA expression of the vtg1 and vtg2 genes in zebrafish. Importantly, the remaining 48 out of 76 potential ER agonists, none of which have previously been reported to have endocrine-disrupting effects or oestrogenic activity, should be of great concern. Our screening results can inform environmental protection goals and play an important role in environmental protection and early warnings to human health.

Data handling and data analysis in metabolomic studies of essential oils using GC-MS

Gas chromatography electron impact ionization mass spectrometry (GC-EI-MS) has been, and remains, the most widely applied analytical technique for metabolomic studies of essential oils. GC-EI-MS analysis of complex samples, such as essential oils, creates a large volume of data. Creating predictive models for such samples and observing patterns within complex data sets presents a significant challenge and requires application of robust data handling and data analysis methods. Accordingly, a wide variety of software and algorithms has been investigated and developed for this purpose over the years. This review provides an overview and summary of that research effort, and attempts to classify and compare different data handling and data analysis procedures that have been reported to-date in the metabolomic study of essential oils using GC-EI-MS.

More Data, Please: Machine Learning to Advance the Multidisciplinary Science of Human Sociochemistry

Communication constitutes the core of human life. A large portion of our everyday social interactions is non-verbal. Of the sensory modalities we use for non-verbal communication, olfaction (i.e., the sense of smell) is often considered the most enigmatic medium. Outside of our awareness, smells provide information about our identity, emotions, gender, mate compatibility, illness, and potentially more. Yet, body odors are astonishingly complex, with their composition being influenced by various factors. Is there a chemical basis of olfactory communication? Can we identify molecules predictive of psychological states and traits? We propose that answering these questions requires integrating two disciplines: psychology and chemistry. This new field, coined sociochemistry, faces new challenges emerging from the sheer amount of factors causing variability in chemical composition of body odorants on the one hand (e.g., diet, hygiene, skin bacteria, hormones, genes), and variability in psychological states and traits on the other (e.g., genes, culture, hormones, internal state, context). In past research, the reality of these high-dimensional data has been reduced in an attempt to isolate unidimensional factors in small, homogenous samples under tightly controlled settings. Here, we propose big data approaches to establish novel links between chemical and psychological data on a large scale from heterogeneous samples in ecologically valid settings. This approach would increase our grip on the way chemical signals non-verbally and subconsciously affect our social lives across contexts.