Evaluating Machine Learning Methods of Analyzing Multiclass Metabolomics

Identification and validation of HOXC6 as a diagnostic biomarker for Ewing sarcoma: insights from machine learning algorithms and in vitro experiments

Introduction

Early diagnosis of Ewing sarcoma (ES) is critical for improving patient prognosis. However, the accurate diagnosis of ES remains challenging, underscoring the need for novel diagnostic biomarkers to enhance diagnostic precision and reliability. This study aimed to identify potential gene expression-based biomarkers for the diagnosis of ES.

Methods

We selected the GSE17679, GSE45544, and GSE68776 datasets from the Gene Expression Omnibus (GEO) database. After correcting for batch effects, we combined ES and normal tissue samples from the GSE17679 and GSE45544 datasets to create a combined cohort. Two-thirds of both the tumor and normal samples from the combined cohort were randomly selected for the training cohort, while the remaining one-third served as the internal validation cohort. Additionally, the GSE68776 dataset was used for external validation. To identify key diagnostic genes, we applied three machine learning algorithms: least absolute shrinkage and selection operator (LASSO), support vector machine recursive feature elimination (SVM-RFE), and random forest (RF).

Results

HOXC6 was identified as a key diagnostic biomarker for ES. It demonstrated strong diagnostic performance across all cohorts, with area under the curve (AUC) values of 0.956 (95% CI: 0.909-0.990) in the training cohort, 0.995 (95% CI: 0.977-1.000) in the internal validation cohort, and 0.966 (95% CI: 0.910-0.999) in the external validation cohort. Functional validation through HOXC6 knockdown in the RD-ES cell line revealed that its suppression significantly inhibited cell proliferation and migration. Furthermore, transcriptome sequencing suggested potential oncogenic mechanisms underlying HOXC6 function.

Discussion

These findings highlight HOXC6 as a promising diagnostic biomarker for ES, demonstrating robust performance across multiple datasets. Additionally, its functional role suggests potential as a therapeutic target.

Exploring Mechanisms and Biomarkers of Breast Cancer Invasion and Migration: An Explainable Gene-Pathway-Compounds Neural Network

BACKGROUNDS

Exploring the molecular features that drive breast cancer invasion and migration remains an important biological and clinical challenge. In recent years, the use of interpretable machine learning models has enhanced our understanding of the underlying mechanisms of disease progression.

METHODS

In this study, we present a novel gene-pathway-compound-related sparse deep neural network (GPC-Net) for investigating breast cancer invasion and migration. The GPC-Net is an interpretable neural network model that utilizes molecular data to predict cancer status. It visually represents genes, pathways, and associated compounds involved in these pathways.

RESULTS

Compared with other modeling methods, GPC-Net demonstrates superior performance. Our research identifies key genes, such as ADCY8, associated with invasive breast cancer and verifies their expression in breast cancer cells. In addition, we conducted a preliminary exploration of several pathways.

CONCLUSION

GPC-Net is among the pioneering deep neural networks that incorporate pathways and compounds, aiming to balance interpretability and performance. It is expected to offer a more convenient approach for future biomedical research.

Benchmark of Data Integration in Single-Cell Proteomics

Single-cell proteomics (SCP) detected based on different technologies always involves batch-specific variations because of differences in sample processing and other potential biases. How to integrate SCP data effectively has become a great challenge. Integration of SCP data not only requires the conservation of true biological variances, but also realizes the removal of unwanted batch effects. In this study, benchmarking analysis of popular data integration methods was conducted to determine the most suitable method for SCP data. To comprehensively evaluate the performance of these integration methods, a novel evaluation system was proposed for integrating SCP data. This evaluation system consists of three objective measures from different perspectives: category (a), the efficacy of correcting batch effects; category (b), the power of conserving biological variances; and category (c), the ability to identify consistent markers. For this comprehensive evaluation, five benchmark data sets under different scenarios (containing substantial proteins, substantial cells, multiple batches, multiple cell types, and unbalanced data) were utilized for selecting the most suitable data integration method. As a result, three methods, ComBat, Scanorama, and Seurat version 3 CCA, were identified as the most recommended methods for integrating SCP data. Overall, this systematic evaluation might provide valuable guidance in choosing the appropriate method for data integration in the SCP.

Imputation of Missing Data in Materials Science through Nearest Neighbors and Iterative Predictions

Missing data in tabular data sets is ubiquitous in statistical analysis, big data analysis, and machine learning studies. Many strategies have been proposed to impute missing data, but their reliability has not been stringently assessed in materials science. Here, we carried out a benchmark test for six imputation strategies: Mean, MissForest, HyperImpute, Gain, Sinkhorn, and a newly proposed MatImpute on seven representative data sets in materials science. The imputation-induced errors (IIEs) were evaluated through the difference between imputed and original values, by root mean square error (RMSE), Wasserstein distance (WD), and a newly introduced metrics data set correlation convergence (DCC), to measure the difference at three aspects for individual data, column-wise distribution, and correlation stability of a data set. MatImpute outperformed the others with the least RMSE and WD and the highest DCC. The IIE increases with the increase of data missing ratio and in the order of missing at random < missing completely at random ≤ missing not at random, considering inherent correlations among missing data. A similar trend was observed for the increase of IIE along the central departure distance in units of the standard deviation, which is consistent with the increase of difficulty from interpolation to extrapolation. Further tests of IIE in regression and classification machine learning predictive models, MatImpute also preserved the highest data recovery fidelity. We released the code of MatImpute to facilitate the construction of high-quality data sets in materials science.

Machine learning based on alcohol drinking-gut microbiota-liver axis in predicting the occurrence of early-stage hepatocellular carcinoma

BACKGROUND

Alcohol drinking and gut microbiota are related to hepatocellular carcinoma (HCC), but the specific relationship between them remains unclear.

AIMS

We aimed to establish the alcohol drinking-gut microbiota-liver axis and develop machine learning (ML) models in predicting the occurrence of early-stage HCC.

METHODS

Two hundred sixty-nine patients with early-stage HCC and 278 controls were recruited. Alcohol drinking-gut microbiota-liver axis was established through the mediation/moderation effect analyses. Eight ML algorithms including Classification and Regression Tree (CART), Gradient Boosting Machine (GBM), K-Nearest Neighbor (KNN), Logistic Regression (LR), Neural Network (NN), Random Forest (RF), Support Vector Machine (SVM), and eXtreme Gradient Boosting (XGBoost) were applied.

RESULTS

A total of 160 pairs of individuals were included for analyses. The mediation effects of Genus_Catenibacterium (P = 0.024), Genus_Tyzzerella_4 (P < 0.001), and Species_Tyzzerella_4 (P = 0.020) were discovered. The moderation effects of Family_Enterococcaceae (OR = 0.741, 95%CI:0.160-0.760, P = 0.017), Family_Leuconostocaceae (OR = 0.793, 95%CI:0.486-3.593, P = 0.010), Genus_Enterococcus (OR = 0.744, 95%CI:0.161-0.753, P = 0.017), Genus_Erysipelatoclostridium (OR = 0.693, 95%CI:0.062-0.672, P = 0.032), Genus_Lactobacillus (OR = 0.655, 95%CI:0.098-0.749, P = 0.011), Species_Enterococcus_faecium (OR = 0.692, 95%CI:0.061-0.673, P = 0.013), and Species_Lactobacillus (OR = 0.653, 95%CI:0.086-0.765, P = 0.014) were uncovered. The predictive power of eight ML models was satisfactory (AUCs:0.855-0.932). The XGBoost model had the best predictive ability (AUC = 0.932).

CONCLUSIONS

ML models based on the alcohol drinking-gut microbiota-liver axis are valuable in predicting the occurrence of early-stage HCC.

Comprehensive Analysis of Bile Medicines Based on UHPLC-QTOF-MSE and Machine Learning

Based on UHPLC-QTOF-MSE analysis and quantized processing, combined with machine learning algorithms, data modeling was carried out to realize digital identification of bear bile powder (BBP), chicken bile powder (CIBP), duck bile powder (DBP), cow bile powder (CBP), sheep bile powder (SBP), pig bile powder (PBP), snake bile powder (SNBP), rabbit bile powder (RBP), and goose bile powder (GBP). First, 173 batches of bile samples were analyzed by UHPLC-QTOF-MSE to obtain the retention time-exact mass (RTEM) data pair to identify bile acid-like chemical components. Then, the data were modeled by combining support vector machine (SVM), random forest (RF), artificial neural network (ANN), gradient boosting (GB), AdaBoost (AB), and Naive Bayes (NB), and the models were evaluated by the parameters of accuracy (Acc), precision (P), and area under the curve (AUC). Finally, the bile medicines were digitally identified based on the optimal model. The results showed that the RF model constructed based on the identified 12 bile acid-like chemical constituents and random forest algorithm is optimal with ACC, P, and AUC > 0.950. In addition, the accuracy of external identification verification of 42 batches of bile medicines detected at different times is 100.0%. So based on UHPLC-QTOF-MSE analysis and combined with the RF algorithm, it can efficiently and accurately realize the digital identification of bile medicines, which can provide reference and assistance for the quality control of bile medicines. In addition, hyodeoxycholic acid, glycohyodeoxycholic acid, and taurochenodeoxycholic acid, and so forth are the most important bile acid constituents for the identification of nine bile medicines.

Creating machine learning models that interpretably link systemic inflammatory index, sex steroid hormones, and dietary antioxidants to identify gout using the SHAP (SHapley Additive exPlanations) method

Background

The relationship between systemic inflammatory index (SII), sex steroid hormones, dietary antioxidants (DA), and gout has not been determined. We aim to develop a reliable and interpretable machine learning (ML) model that links SII, sex steroid hormones, and DA to gout identification.

Methods

The dataset we used to study the relationship between SII, sex steroid hormones, DA, and gout was from the National Health and Nutrition Examination Survey (NHANES). Six ML models were developed to identify gout by SII, sex steroid hormones, and DA. The seven performance discriminative features of each model were summarized, and the eXtreme Gradient Boosting (XGBoost) model with the best overall performance was selected to identify gout. We used the SHapley Additive exPlanation (SHAP) method to explain the XGBoost model and its decision-making process.

Results

An initial survey of 20,146 participants resulted in 8,550 being included in the study. Selecting the best performing XGBoost model associated with SII, sex steroid hormones, and DA to identify gout (male: AUC: 0.795, 95% CI: 0.746- 0.843, accuracy: 98.7%; female: AUC: 0.822, 95% CI: 0.754- 0.883, accuracy: 99.2%). In the male group, The SHAP values showed that the lower feature values of lutein + zeaxanthin (LZ), vitamin C (VitC), lycopene, zinc, total testosterone (TT), vitamin E (VitE), and vitamin A (VitA), the greater the positive effect on the model output. In the female group, SHAP values showed that lower feature values of E2, zinc, lycopene, LZ, TT, and selenium had a greater positive effect on model output.

Conclusion

The interpretable XGBoost model demonstrated accuracy, efficiency, and robustness in identifying associations between SII, sex steroid hormones, DA, and gout in participants. Decreased TT in males and decreased E2 in females may be associated with gout, and increased DA intake and decreased SII may reduce the potential risk of gout.

Advancements in CHO metabolomics: techniques, current state and evolving methodologies

Background: Investigating the metabolic behaviour of different cellular phenotypes, i.e., good/bad grower and/or producer, in production culture is important to identify the key metabolite(s)/pathway(s) that regulate cell growth and/or recombinant protein production to improve the overall yield. Currently, LC-MS, GC-MS and NMR are the most used and advanced technologies for investigating the metabolome. Although contributed significantly in the domain, each technique has its own biasness towards specific metabolites or class of metabolites due to various reasons including variability in the concept of working, sample preparation, metabolite-extraction methods, metabolite identification tools, and databases. As a result, the application of appropriate analytical technique(s) is very critical. Purpose and scope: This review provides a state-of-the-art technological insights and overview of metabolic mechanisms involved in regulation of cell growth and/or recombinant protein production for improving yield from CHO cultures. Summary and conclusion: In this review, the advancements in CHO metabolomics over the last 10 years are traced based on a bibliometric analysis of previous publications and discussed. With the technical advancement in the domain of LC-MS, GC-MS and NMR, metabolites of glycolytic and nucleotide biosynthesis pathway (glucose, fructose, pyruvate and phenylalanine, threonine, tryptophan, arginine, valine, asparagine, and serine, etc.) were observed to be upregulated in exponential-phase thereby potentially associated with cell growth regulation, whereas metabolites/intermediates of TCA, oxidative phosphorylation (aspartate, glutamate, succinate, malate, fumarate and citrate), intracellular NAD+/NADH ratio, and glutathione metabolic pathways were observed to be upregulated in stationary-phase and hence potentially associated with increased cell-specific productivity in CHO bioprocess. Moreover, each of technique has its own bias towards metabolite identification, indicating their complementarity, along with a number of critical gaps in the CHO metabolomics pipeline and hence first time discussed here to identify their potential remedies. This knowledge may help in future study designs to improve the metabolomic coverage facilitating identification of the metabolites/pathways which might get missed otherwise and explore the full potential of metabolomics for improving the CHO bioprocess performances.

[Identification of Osteoarthritis Inflamm-Aging Biomarkers by Integrating Bioinformatic Analysis and Machine Learning Strategies and the Clinical Validation]

Objective

To identify inflamm-aging related biomarkers in osteoarthritis (OA).

Methods

Microarray gene profiles of young and aging OA patients were obtained from the Gene Expression Omnibus (GEO) database and aging-related genes (ARGs) were obtained from the Human Aging Genome Resource (HAGR) database. The differentially expressed genes of young OA and older OA patients were screened and then intersected with ARGs to obtain the aging-related genes of OA. Enrichment analysis was performed to reveal the potential mechanisms of aging-related markers in OA. Three machine learning methods were used to identify core senescence markers of OA and the receiver operating characteristic (ROC) curve was used to assess their diagnostic performance. Peripheral blood mononuclear cells were collected from clinical OA patients to verify the expression of senescence-associated secretory phenotype (SASP) factors and senescence markers.

Results

A total of 45 senescence-related markers were obtained, which were mainly involved in the regulation of cellular senescence, the cell cycle, inflammatory response, etc. Through the screening with the three machine learning methods, 5 core senescence biomarkers, including FOXO3, MCL1, SIRT3, STAG1, and S100A13, were obtained. A total of 20 cases of normal controls and 40 cases of OA patients, including 20 cases in the young patient group and 20 in the elderly patient group, were enrolled. Compared with those of the young patient group, C-reactive protein (CRP), interleukin (IL)-6, and IL-1β levels increased and IL-4 levels decreased in the elderly OA patient group (P<0.01); FOXO3, MCL1, and SIRT3 mRNA expression decreased and STAG1 and S100A13 mRNA expression increased (P<0.01). Pearson correlation analysis demonstrated that the selected markers were associated with some indicators, including erythrocyte sedimentation rate (ESR), IL-1β, IL-4, CRP, and IL-6. The area under the ROC curve of the 5 core aging genes was always greater than 0.8 and the C-index of the calibration curve in the nomogram prediction model was 0.755, which suggested the good calibration ability of the model.

Conclusion

FOXO3, MCL1, SIRT3, STAG1, and S100A13 may serve as novel diagnostic biomolecular markers and potential therapeutic targets for OA inflamm-aging.