An approach for feature selection with data modelling in LC-MS metabolomics

Analysis of the lipidomic profile of vegetable oils and animal fats and changes during aging by UPLC-Q-exactive orbitrap mass spectrometry

Vegetable oil and animal fat residues are common evidence in the cases of homicide, arson, theft, and other crimes. However, the lipid composition and content changes during aging on complex carriers remain unclear. Therefore, this study dynamically monitored the lipid composition and content changes during aging of 13 different types of vegetable oils and animal fats on five different carriers using the UPLC-Q-Exactive Orbitrap MS method. A total of 6 subclasses of 93 lipids including lysophosphatidylcholine (2 species), phosphatidylcholine (2 species), diglyceride (5 species), triglyceride (81 species), acylGlcCampesterol ester (2 species), and acylGlcSitosterol ester (1 species), were first identified in fresh vegetable oils and animal fats. By comparing the LC-MS/MS chromatograms of fresh vegetable oils and animal fats, it was found that there were significant differences between the chromatograms of vegetable oils and animal fats, but it was difficult to distinguish between the chromatograms of vegetable oils or animal fats. After aging at 60 °C for 200 days, there was a significant decrease in the content of diglyceride, triglyceride, acylGlcCampesterol ester, and acylGlcSitosterol ester, while the content of lysophosphatidylcholine and phosphatidylcholine initially increased and then decreased. Furthermore, statistical analysis of lipid differences between vegetable oils and animal fats was performed using cluster heat maps, volcanic maps, PCA, and OPLS-DA. On average, 33 significantly different lipids were screened (VIP > 1, p < 0.05), which could serve as potential biomarkers for distinguishing vegetable oils and animal fats. It was found that the potential biomarkers still existed during aging of vegetable oils and animal fats (100 and 200 days). This research provides important reference information for the identification of vegetable oil and animal fat residues in complex carriers at crime scenes.

Union With Recursive Feature Elimination: A Feature Selection Framework to Improve the Classification Performance of Multicategory Causes of Death in Colorectal Cancer

Despite the use of machine learning tools, it is challenging to properly model cause-specific deaths in colorectal cancer (CRC) patients and choose appropriate treatments. Here, we propose an interesting feature selection framework, namely union with recursive feature elimination (U-RFE), to select the union feature sets that are crucial in CRC progression-specific mortality using The Cancer Genome Atlas (TCGA) dataset. Based on the union feature sets, we compared the performance of 5 classification algorithms, including logistic regression (LR), support vector machines (SVM), random forest (RF), eXtreme gradient boosting (XGBoost), and Stacking, to identify the best model for classifying 4-category deaths. In the first stage of U-RFE, LR, SVM, and RF were used as base estimators to obtain subsets containing the same number of features but not exactly the same specific features. Union analysis of the subsets was then performed to determine the final union feature set, effectively combining the advantages of different algorithms. We found that the U-RFE framework could improve various models' performance. Stacking outperformed LR, SVM, RF, and XGBoost in most scenarios. When the target feature number of the RFE was set to 50 and the union feature set contained 298 deterministic features, the Stacking model achieved F1_weighted, Recall_weighted, Precision_weighted, Accuracy, and Matthews correlation coefficient of 0.851, 0.864, 0.854, 0.864, and 0.717, respectively. The performance of the minority categories was also significantly improved. Therefore, this recursive feature elimination-based approach of feature selection improves performances of classifying CRC deaths using clinical and omics data or those using other data with high feature redundancy and imbalance.

An Iterative Curve-Fitting Baseline Correction Method for Raman Spectra Driven by Neural Network

Baseline correction is a vital part of spectral preprocessing, especially for Raman spectra. Iterative polynomial fitting is an easy but less accurate way to find baselines compared to other methods such as wavelet transform and penalized least squares (PLS) methods. The polynomial fitting methods can also get distorted results in certain conditions. In this paper, a neural network model for detecting the trend of the baseline was proposed to improve the correction accuracy of the fitting methods. The model selects the function basis according to the baseline trend instead of using a fixed polynomial function to match the baseline for a more precise fit. We also propose a way to generate simulation data, these data can be used to train the neural network model. The model provides reliable results for real spectral data with noise. Our method provides a new idea to correct the baseline with a strange shape. In addition, we examine the limitations of conventional iterative polynomial fitting, adaptive iteratively reweighted PLS and explain why our approach surpasses these methods.

A Computational Approach to Identification of Candidate Biomarkers in High-Dimensional Molecular Data

Complex high-dimensional datasets that are challenging to analyze are frequently produced through ‘-omics’ profiling. Typically, these datasets contain more genomic features than samples, limiting the use of multivariable statistical and machine learning-based approaches to analysis. Therefore, effective alternative approaches are urgently needed to identify features-of-interest in ‘-omics’ data. In this study, we present the molecular feature selection tool, a novel, ensemble-based, feature selection application for identifying candidate biomarkers in ‘-omics’ data. As proof-of-principle, we applied the molecular feature selection tool to identify a small set of immune-related genes as potential biomarkers of three prostate adenocarcinoma subtypes. Furthermore, we tested the selected genes in a model to classify the three subtypes and compared the results to models built using all genes and all differentially expressed genes. Genes identified with the molecular feature selection tool performed better than the other models in this study in all comparison metrics: accuracy, precision, recall, and F1-score using a significantly smaller set of genes. In addition, we developed a simple graphical user interface for the molecular feature selection tool, which is available for free download. This user-friendly interface is a valuable tool for the identification of potential biomarkers in gene expression datasets and is an asset for biomarker discovery studies.

Computational Workflow to Study the Diversity of Secondary Metabolites in Fourteen Different Isatis Species

The screening of real features among thousands of ions remains a great challenge in the study of metabolomics. In this research, a workflow designed based on the MetaboFR tool and “feature-rating” rule was developed to screen the real features in large-scale data analyses. Seventy-four reference standards were used to test the feasibility, with 83.21% of real features being obtained after MetaboFR processing. Moreover, the full workflow was applied for systematic characterization of 14 species of the genus Isatis, with the result that 87.72% of real features were retained and 69.19% of the in-source fragments were removed. To gain insights into metabolite diversity within this plant family, 1697 real features were tentatively identified, including lipids, phenylpropanoids, organic acids, indole derivatives, etc. Indole derivatives were demonstrated to be the best chemical markers with which to differentiate different species. The rare existence of indole derivatives in Isatis cappadocica (cap) and Isatis cappadocica subsp. Steveniana (capS) indicates that the biosynthesis of indole derivatives could play a key role in driving the chemical diversity and evolution of genus Isatis. Our workflow provides the foundations for the exploration of real features in metabolomics, and has the potential to reveal the chemical composition and marker metabolites of secondary metabolites in plant fields.

Untargeted and targeted analysis of sarin poisoning biomarkers in rat urine by liquid chromatography and tandem mass spectrometry

Chemical warfare agents continue to pose a real threat to humanity, despite their prohibition under the Chemical Weapons Convention. Sarin is one of the most toxic and lethal representatives of nerve agents. The methodology for the targeted analysis of known sarin metabolites has reached great heights, but little attention has been paid to the untargeted analysis of biological samples of victims exposed to this deadly poisonous substance. At present, the development of computational and statistical methods of analysis offers great opportunities for finding new metabolites or understanding the mechanisms of action or effect of toxic substances on the organism. This study presents the targeted LC-MS/MS determination of methylphosphonic acid and isopropyl methylphosphonic acid in the urine of rats exposed to a non-lethal dose of sarin, as well as the untarget urine analysis by LC-HRMS. Targeted analysis of polar acidic sarin metabolites was performed on a mixed-mode reversed-phase anion-exchange column, and untargeted analysis on a conventional reversed-phase C18 column. Isopropyl methylphosphonic acid was detected and quantified within 5 days after subcutaneous injection of sarin at a dose of 1/4 LD₅₀. A combination of generalized additive mixed models and dose-response analysis with database searches using accurate mass of precursor ions and corresponding MS/MS spectra enabled us to propose new six potential biomarkers of biological response to exposure. The results confirm the well-known fact that sarin poisoning has a significant impact on the victims' metabolome, with inhibition of acetylcholinesterase being just the first step and trigger of the complex toxicodynamic response.