Application of high-dimensional feature selection: evaluation for genomic prediction in man

Exploring genomic feature selection: A comparative analysis of GWAS and machine learning algorithms in a large-scale soybean dataset

The surge in high-throughput technologies has empowered the acquisition of vast genomic datasets, prompting the search for genetic markers and biomarkers relevant to complex traits. However, grappling with the inherent complexities of high dimensionality and sparsity within these datasets poses formidable hurdles. The immense number of features and their potential redundancy demand efficient strategies for extracting pertinent information and identifying significant markers. Feature selection is important in large genomic data as it helps in enhancing interpretability and computational efficiency. This study focuses on addressing these challenges through a comprehensive investigation into genomic feature selection methodologies, employing a rich soybean (Glycine max L. Merr.) dataset comprising 966 lines with over 5.5 million single nucleotide polymorphisms. Emphasizing the "small n large p" dilemma prevalent in contemporary genomic studies, we compared the efficacy of traditional genome-wide association studies (GWAS) with two prominent machine learning tools, random forest and extreme gradient boosting, in pinpointing predictive features. Utilizing the expansive soybean dataset, we assessed the performance of these methodologies in selecting features that optimize predictive modeling for various phenotypes. By constructing predictive models based on the selected features, we ascertain the comparative prediction accuracies, thereby illuminating the strengths and limitations of these feature selection methodologies in the realm of genomic data analysis.

Identification of Psychological Treatment Dropout Predictors Using Machine Learning Models on Italian Patients Living with Overweight and Obesity Ineligible for Bariatric Surgery

According to the main international guidelines, patients with obesity and psychiatric/psychological disorders who cannot be addressed to surgery are recommended to follow a nutritional approach and a psychological treatment. A total of 94 patients (T0) completed a battery of self-report measures: Symptom Checklist-90-Revised (SCL-90-R), Barratt Impulsiveness Scale-11 (BIS-11), Binge-Eating Scale (BES), Obesity-Related Well-Being Questionnaire-97 (ORWELL-97), and Minnesota Multiphasic Personality Inventory-2 (MMPI-2). Then, twelve sessions of a brief psychodynamic psychotherapy were delivered, which was followed by the participants completing the follow-up evaluation (T1). Two groups of patients were identified: Group 1 (n = 65), who fully completed the assessment in both T0 and T1; and Group 2-dropout (n = 29), who fulfilled the assessment only at T0 and not at T1. Machine learning models were implemented to investigate which variables were most associated with treatment failure. The classification tree model identified patients who were dropping out of treatment with an accuracy of about 80% by considering two variables: the MMPI-2 Correction (K) scale and the SCL-90-R Phobic Anxiety (PHOB) scale. Given the limited number of studies on this topic, the present results highlight the importance of considering the patient's level of adaptation and the social context in which they are integrated in treatment planning. Cautionary notes, implications, and future directions are discussed.

Feature engineering of environmental covariates improves plant genomic-enabled prediction

Introduction

Because Genomic selection (GS) is a predictive methodology, it needs to guarantee high-prediction accuracies for practical implementations. However, since many factors affect the prediction performance of this methodology, its practical implementation still needs to be improved in many breeding programs. For this reason, many strategies have been explored to improve the prediction performance of this methodology.

Methods

When environmental covariates are incorporated as inputs in the genomic prediction models, this information only sometimes helps increase prediction performance. For this reason, this investigation explores the use of feature engineering on the environmental covariates to enhance the prediction performance of genomic prediction models.

Results and discussion

We found that across data sets, feature engineering helps reduce prediction error regarding only the inclusion of the environmental covariates without feature engineering by 761.625% across predictors. These results are very promising regarding the potential of feature engineering to enhance prediction accuracy. However, since a significant gain in prediction accuracy was observed in only some data sets, further research is required to guarantee a robust feature engineering strategy to incorporate the environmental covariates.

Reservoir temperature prediction based on characterization of water chemistry data-case study of western Anatolia, Turkey

Reservoir temperature estimation is crucial for geothermal studies, but traditional methods are complex and uncertain. To address this, we collected 83 sets of water chemistry and reservoir temperature data and applied four machine learning algorithms. These models considered various input factors and underwent data preprocessing steps like null value imputation, normalization, and Pearson coefficient calculation. Cross-validation addressed data volume issues, and performance metrics were used for model evaluation. The results revealed that our machine learning models outperformed traditional fluid geothermometers. All machine learning models surpassed traditional methods. The XGBoost model, based on the F-3 combination, demonstrated the best prediction accuracy with an R2 of 0.9732, while the Bayesian ridge regression model using the F-4 combination had the lowest performance with an R2 of 0.8302. This study highlights the potential of machine learning for accurate reservoir temperature prediction, offering geothermal professionals a reliable tool for model selection and advancing our understanding of geothermal resources.

Enhancing prediction accuracy of coronary artery disease through machine learning-driven genomic variant selection

Machine learning (ML) methods are increasingly becoming crucial in genome-wide association studies for identifying key genetic variants or SNPs that statistical methods might overlook. Statistical methods predominantly identify SNPs with notable effect sizes by conducting association tests on individual genetic variants, one at a time, to determine their relationship with the target phenotype. These genetic variants are then used to create polygenic risk scores (PRSs), estimating an individual's genetic risk for complex diseases like cancer or cardiovascular disorders. Unlike traditional methods, ML algorithms can identify groups of low-risk genetic variants that improve prediction accuracy when combined in a mathematical model. However, the application of ML strategies requires addressing the feature selection challenge to prevent overfitting. Moreover, ensuring the ML model depends on a concise set of genomic variants enhances its clinical applicability, where testing is feasible for only a limited number of SNPs. In this study, we introduce a robust pipeline that applies ML algorithms in combination with feature selection (ML-FS algorithms), aimed at identifying the most significant genomic variants associated with the coronary artery disease (CAD) phenotype. The proposed computational approach was tested on individuals from the UK Biobank, differentiating between CAD and non-CAD individuals within this extensive cohort, and benchmarked against standard PRS-based methodologies like LDpred2 and Lassosum. Our strategy incorporates cross-validation to ensure a more robust evaluation of genomic variant-based prediction models. This method is commonly applied in machine learning strategies but has often been neglected in previous studies assessing the predictive performance of polygenic risk scores. Our results demonstrate that the ML-FS algorithm can identify panels with as few as 50 genetic markers that can achieve approximately 80% accuracy when used in combination with known risk factors. The modest increase in accuracy over PRS performances is noteworthy, especially considering that PRS models incorporate a substantially larger number of genetic variants. This extensive variant selection can pose practical challenges in clinical settings. Additionally, the proposed approach revealed novel CAD-genetic variant associations.

Machine learning for microbiologists

Machine learning is increasingly important in microbiology where it is used for tasks such as predicting antibiotic resistance and associating human microbiome features with complex host diseases. The applications in microbiology are quickly expanding and the machine learning tools frequently used in basic and clinical research range from classification and regression to clustering and dimensionality reduction. In this Review, we examine the main machine learning concepts, tasks and applications that are relevant for experimental and clinical microbiologists. We provide the minimal toolbox for a microbiologist to be able to understand, interpret and use machine learning in their experimental and translational activities.

Genomic selection in plant breeding: Key factors shaping two decades of progress

Genomic selection, the application of genomic prediction (GP) models to select candidate individuals, has significantly advanced in the past two decades, effectively accelerating genetic gains in plant breeding. This article provides a holistic overview of key factors that have influenced GP in plant breeding during this period. We delved into the pivotal roles of training population size and genetic diversity, and their relationship with the breeding population, in determining GP accuracy. Special emphasis was placed on optimizing training population size. We explored its benefits and the associated diminishing returns beyond an optimum size. This was done while considering the balance between resource allocation and maximizing prediction accuracy through current optimization algorithms. The density and distribution of single-nucleotide polymorphisms, level of linkage disequilibrium, genetic complexity, trait heritability, statistical machine-learning methods, and non-additive effects are the other vital factors. Using wheat, maize, and potato as examples, we summarize the effect of these factors on the accuracy of GP for various traits. The search for high accuracy in GP-theoretically reaching one when using the Pearson's correlation as a metric-is an active research area as yet far from optimal for various traits. We hypothesize that with ultra-high sizes of genotypic and phenotypic datasets, effective training population optimization methods and support from other omics approaches (transcriptomics, metabolomics and proteomics) coupled with deep-learning algorithms could overcome the boundaries of current limitations to achieve the highest possible prediction accuracy, making genomic selection an effective tool in plant breeding.

Selective Genotyping and Phenotyping for Optimization of Genomic Prediction Models for Populations with Different Diversity

To overcome the different challenges to food security caused by a growing population and climate change, soybean (Glycine max (L.) Merr.) breeders are creating novel cultivars that have the potential to improve productivity while maintaining environmental sustainability. Genomic selection (GS) is an advanced approach that may accelerate the rate of genetic gain in breeding using genome-wide molecular markers. The accuracy of genomic selection can be affected by trait architecture and heritability, marker density, linkage disequilibrium, statistical models, and training set. The selection of a minimal and optimal marker set with high prediction accuracy can lower genotyping costs, computational time, and multicollinearity. Selective phenotyping could reduce the number of genotypes tested in the field while preserving the genetic diversity of the initial population. This study aimed to evaluate different methods of selective genotyping and phenotyping on the accuracy of genomic prediction for soybean yield. The evaluation was performed on three populations: recombinant inbred lines, multifamily diverse lines, and germplasm collection. Strategies adopted for marker selection were as follows: SNP (single nucleotide polymorphism) pruning, estimation of marker effects, randomly selected markers, and genome-wide association study. Reduction of the number of genotypes was performed by selecting a core set from the initial population based on marker data, yet maintaining the original population's genetic diversity. Prediction ability using all markers and genotypes was different among examined populations. The subsets obtained by the model-based strategy can be considered the most suitable for marker selection for all populations. The selective phenotyping based on makers in all cases had higher values of prediction ability compared to minimal values of prediction ability of multiple cycles of random selection, with the highest values of prediction obtained using AN approach and 75% population size. The obtained results indicate that selective genotyping and phenotyping hold great potential and can be integrated as tools for improving or retaining selection accuracy by reducing genotyping or phenotyping costs for genomic selection.

Deep self-supervised machine learning algorithms with a novel feature elimination and selection approaches for blood test-based multi-dimensional health risks classification

BACKGROUND

Blood test is extensively performed for screening, diagnoses and surveillance purposes. Although it is possible to automatically evaluate the raw blood test data with the advanced deep self-supervised machine learning approaches, it has not been profoundly investigated and implemented yet.

RESULTS

This paper proposes deep machine learning algorithms with multi-dimensional adaptive feature elimination, self-feature weighting and novel feature selection approaches. To classify the health risks based on the processed data with the deep layers, four machine learning algorithms having various properties from being utterly model free to gradient driven are modified.

CONCLUSIONS

The results show that the proposed deep machine learning algorithms can remove the unnecessary features, assign self-importance weights, selects their most informative ones and classify the health risks automatically from the worst-case low to worst-case high values.

Effects of lifestyle behaviours and depressed mood on sleep quality in young adults. A machine learning approach

BACKGROUND

Modern lifestyles may lead to high stress levels, frequently associated with mood disorders (e.g. depressed mood) and sleep disturbance. The objective of this study was to develop a machine learning model aimed at identifying risk factors for developing poor sleep quality in young adults.

MATERIAL AND METHODS

The sample consisted of 383 college-aged students (mean age ± SD: 21 ± 1 years; 61% males). Sleep quality, mood state, physical activity, number of sitting hours, and smartphone use were measured.

RESULTS

A decision tree algorithm distinguished participants' sleep quality with 74% accuracy using a combination of four features: depressed mood, physical activity, sitting time, and vigour. Together with depressed mood, both physical activity (>6432 metabolic equivalent tasks -METs- per week) and sedentary behaviour (sitting time greater than 7 h/day) were the primary features that could differentiate those with poor sleep quality from those with good sleep quality.

CONCLUSIONS

We provided a decision tree model with a sensitivity of 90.7% and a specificity of 54.3%, with an AUC of 0.725. These findings could promote improvements in prevention strategies and contribute to the development of meaningful and evidence-based intervention programs.

Improving algal bloom detection using spectroscopic analysis and machine learning: A case study in a large artificial reservoir, South Korea

The prediction of algal blooms using traditional water quality indicators is expensive, labor-intensive, and time-consuming, making it challenging to meet the critical requirement of timely monitoring for prompt management. Using optical measures for forecasting algal blooms is a feasible and useful method to overcome these problems. This study explores the potential application of optical measures to enhance algal bloom prediction in terms of prediction accuracy and workload reduction, aided by machine learning (ML) models. Compared to absorption-derived parameters, commonly used fluorescence indices such as the fluorescence index (FI), humification index (HIX), biological index (BIX), and protein-like component improved the prediction accuracy. However, the prediction accuracy was decreased when all optical indices were considered for computation due to increased noise and uncertainty in the models. With the exception of chemical oxygen demand (COD), this study successfully replaced biochemical oxygen demand (BOD), dissolved organic carbon (DOC), and nutrients with selected fluorescence indices, demonstrating relatively analogous performance in either training or testing data, with consistent and good coefficient of determination (R2) values of approximately 0.85 and 0.74, respectively. Among all models considered, ensemble learning models consistently outperformed conventional regression models and artificial neural networks (ANNs). However, there was a trade-off between accuracy and computation efficiency among the ensemble learning models (i.e., Stacking and XGBoost) for algal bloom prediction. Our study offers a glimpse of the potential application of spectroscopic measures to improve accuracy and efficiency in algal bloom prediction, but further work should be carried out in other water bodies to further validate our proposed hypothesis.

Exploring the potential of incremental feature selection to improve genomic prediction accuracy

BACKGROUND

The ever-increasing availability of high-density genomic markers in the form of single nucleotide polymorphisms (SNPs) enables genomic prediction, i.e. the inference of phenotypes based solely on genomic data, in the field of animal and plant breeding, where it has become an important tool. However, given the limited number of individuals, the abundance of variables (SNPs) can reduce the accuracy of prediction models due to overfitting or irrelevant SNPs. Feature selection can help to reduce the number of irrelevant SNPs and increase the model performance. In this study, we investigated an incremental feature selection approach based on ranking the SNPs according to the results of a genome-wide association study that we combined with random forest as a prediction model, and we applied it on several animal and plant datasets.

RESULTS

Applying our approach to different datasets yielded a wide range of outcomes, i.e. from a substantial increase in prediction accuracy in a few cases to minor improvements when only a fraction of the available SNPs were used. Compared with models using all available SNPs, our approach was able to achieve comparable performances with a considerably reduced number of SNPs in several cases. Our approach showcased state-of-the-art efficiency and performance while having a faster computation time.

CONCLUSIONS

The results of our study suggest that our incremental feature selection approach has the potential to improve prediction accuracy substantially. However, this gain seems to depend on the genomic data used. Even for datasets where the number of markers is smaller than the number of individuals, feature selection may still increase the performance of the genomic prediction. Our approach is implemented in R and is available at https://github.com/FelixHeinrich/GP_with_IFS/ .

Integrating genome-wide association study into genomic selection for the prediction of agronomic traits in rice (Oryza sativa L.)

Accurately identifying varieties with targeted agronomic traits was thought to contribute to genetic selection and accelerate rice breeding progress. Genomic selection (GS) is a promising technique that uses markers covering the whole genome to predict the genomic-estimated breeding values (GEBV), with the ability to select before phenotypes are measured. To choose the appropriate GS models for breeding work, we analyzed the predictability of nine agronomic traits measured from a population of 459 diverse rice varieties. By the comparison of eight representative GS models, we found that the prediction accuracies ranged from 0.407 to 0.896, with reproducing kernel Hilbert space (RKHS) having the highest predictive ability in most traits. Further results demonstrated the predictivity of GS is altered by several factors. Moreover, we assessed the method of integrating genome-wide association study (GWAS) into various GS models. The predictabilities of GS combined peak-associated markers generated from six different GWAS models were significantly different; a recommendation of Mixed Linear Model (MLM)-RKHS was given for the GWAS-GS-integrated prediction. Finally, based on the above result, we experimented with applying the P-values obtained from optimal GWAS models into ridge regression best linear unbiased prediction (rrBLUP), which benefited the low predictive traits in rice.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11032-023-01423-y.

Improving efficiency of fitting Cox proportional hazards models for time-to-event outcomes in genome-wide association studies (GWAS)

Summary

Technologies identifying single nucleotide polymorphisms (SNPs) in DNA sequencing yield an avalanche of data requiring analysis and interpretation. Standard methods may require many weeks of processing time. The use of statistical methods requiring data sorting, matrix inversions of a high-dimension and replication in subsets of the data on multiple outcomes exacerbate these times.A method which reduces the computational time in problems with time-to-event outcomes and hundreds of thousands/millions of SNPs using Cox-Snell residuals after fitting the Cox proportional hazards model (PH) to a fixed set of concomitant variables is proposed. This yields coefficients for SNP effect from a Cox-Snell adjusted Poisson model and shows a high concordance to the adjusted PH model.The method is illustrated with a sample of 10 000 SNPs from a genome-wide association study in a diabetic population. The gain in processing efficiency using the proposed method based on Poisson modelling can be as high as 62%. This could result in saving of over three weeks processing time if 5 million SNPs require analysis. The method involves only a single predictor variable (SNP), offering a simpler, computationally more stable approach to examining and identifying SNP patterns associated with the outcome(s) allowing for a faster development of genetic signatures. Use of deviance residuals from the PH model to screen SNPs demonstrates a large discordance rate at a 0.2% threshold of concordance. This rate is 15 times larger than that based on the Cox-Snell residuals from the Cox-Snell adjusted Poisson model.

Availability and implementation

The method is simple to implement as the procedures are available in most statistical packges. The approach involves obtaining Cox-Snell residuals from a PH model, to a binary time-to-event outcome, for factors which need to be common when assessing each SNP. Each SNP is then fitted as a predictor to the outcome of interest using a Poisson model with the Cox-Snell as the exposure variable.

A genomic estimated breeding value-assisted reduction method of single nucleotide polymorphism sets: a novel approach for determining the cutoff thresholds in genome-wide association studies and best linear unbiased prediction

Traditionally, the p-value is the criterion for the cutoff threshold to determine significant markers in genome-wide association studies (GWASs). Choosing the best subset of markers for the best linear unbiased prediction (BLUP) for improved prediction ability (PA) has become an interesting issue. However, when dealing with many traits having the same marker information, the p-values' themselves cannot be used as an obvious solution for having a confidence in GWAS and BLUP. We thus suggest a genomic estimated breeding value-assisted reduction method of the single nucleotide polymorphism (SNP) set (GARS) to address these difficulties. GARS is a BLUP-based SNP set decision presentation. The samples were Landrace pigs and the traits used were back fat thickness (BF) and daily weight gain (DWG). The prediction abilities (PAs) for BF and DWG for the entire SNP set were 0.8 and 0.8, respectively. By using the correlation between genomic estimated breeding values (GEBVs) and phenotypic values, selecting the cutoff threshold in GWAS and the best SNP subsets in BLUP was plausible as defined by GARS method. 6,000 SNPs in BF and 4,000 SNPs in DWG were considered as adequate thresholds. Gene Ontology (GO) analysis using the GARS results of the BF indicated neuron projection development as the notable GO term, whereas for the DWG, the main GO terms were nervous system development and cell adhesion.

Identification of Novel Biomarkers for Response to Preoperative Chemoradiation in Locally Advanced Rectal Cancer with Genetic Algorithm-Based Gene Selection

BACKGROUND

The conventional treatment for patients with locally advanced colorectal tumors is preoperative chemo-radiotherapy (PCRT) preceding surgery. This treatment strategy has some long-term side effects, and some patients do not respond to it. Therefore, an evaluation of biomarkers that may help predict patients' response to PCRT is essential.

METHODS

We took advantage of genetic algorithm to search the space of possible combinations of features to choose subsets of genes that would yield convenient performance in differentiating PCRT responders from non-responders using a logistic regression model as our classifier.

RESULTS

We developed two gene signatures; first, to achieve the maximum prediction accuracy, the algorithm yielded 39 genes, and then, aiming to reduce the feature numbers as much as possible (while maintaining acceptable performance), a 5-gene signature was chosen. The performance of the two gene signatures was (accuracy = 0.97 and 0.81, sensitivity = 0.96 and 0.83, and specificity = 86 and 0.77) using a logistic regression classifier. Through analyzing bias and variance decomposition of the model error, we further investigated the involved genes by discovering and validating another 28-gene signature which possibly points towards two different sub-systems involved in the response of the patients to treatment.

CONCLUSIONS

Using genetic algorithm as our gene selection method, we have identified two groups of genes that can differentiate PCRT responders from non-responders in patients of the studied dataset with considerable performance.

IMPACT

After passing standard requirements, our gene signatures may be applicable as a robust and effective PCRT response prediction tool for colorectal cancer patients in clinical settings and may also help future studies aiming to further investigate involved pathways gain a clearer picture for the course of their research.

Identification of Potent Small-Molecule PCSK9 Inhibitors Based on Quantitative Structure-Activity Relationship, Pharmacophore Modeling, and Molecular Docking Procedure

The leading cause of atherosclerotic cardiovascular disease (ASCVD) is elevated low-density lipoprotein cholesterol (LDL-C). Proprotein convertase subtilisin/kexin type 9 (PCSK9) attaches to the domain of LDL receptor (LDLR), diminishing LDL-C influx and LDLR cell surface presentation in hepatocytes, resulting in higher circulating LDL-C levels. PCSK9 dysfunction has been linked to lower levels of plasma LDLC and a decreased risk of coronary heart disease (CHD). Herein, using virtual screening tools, we aimed to identify a potent small-molecule PCSK9 inhibitor in compounds that are currently being studied in clinical trials. We first performed chemical absorption, distribution, metabolism, excretion, and toxicity (ADMET) filtering of 9800 clinical trial compounds obtained from the ZINC 15 database using Lipinski's rule of 5 and achieved 3853 compounds. Two-dimensional (2D) quantitative structure-activity relationship (QSAR) was initiated by computing molecular descriptors and selecting important descriptors of 23 PCSK9 inhibitors. Multivariate calibration was performed with the partial least square regression (PLS) method with 18 compounds for training to design the QSAR model and 5 compounds for the test set to assess the model. The best latent variables (LV) (LV=6) with the lowest value of Root-Mean-Square Error of Cross-Validation (RMSECV) of 0.48 and leave-one-out cross-validation correlation coefficient (R2CV) = 0.83 were obtained for the QSAR model. The low RMSEC (0.21) with high R²cal (0.966) indicates the probability of fit between the experimental data and the calibration model. Using QSAR analysis of 3853 compounds, 2635 had a pIC50<1 and were considered for pharmacophore screening. The PHASE module (a complete package for pharmacophore modeling) designed the pharmacophore hypothesis through multiple ligands. The top 14 compounds (pIC50>1) were defined as active, whereas 9 (pIC50<1) were considered as an inactive set. Three five-point pharmacophore hypotheses achieved the highest score: DHHRR1, DHHRR2, and DHRRR1. The highest and best model with survival scores (5.365) was DHHRR1, comprising 1 hydrogen donor (D), 2 hydrophobic groups (H), and 2 rings of aromatic (R) features. We selected the molecules with a higher 1.5 fitness score (257 compounds) in pharmacophore screening (DHHRR1) for molecular docking screening. Molecular docking indicates that ZINC000051951669, with a binding affinity: of -13.2 kcal/mol and 2 H-bonds, has the highest binding to the PCSK9 protein. ZINC000011726230 with energy binding: -11.4 kcal/mol and 3 H-bonds, ZINC000068248147 with binding affinity: -10.7 kcal/mol and 1 H-bond, ZINC000029134440 with a binding affinity: -10.6 kcal/mol and 4 H-bonds were ranked next, respectively. To conclude, the archived molecules identified as inhibitory PCSK9 candidates, and especially ZINC000051951669 may therefore significantly inhibit PCSK9 and should be considered in the newly designed trials.

FSF-GA: A Feature Selection Framework for Phenotype Prediction Using Genetic Algorithms

(1) Background: Phenotype prediction is a pivotal task in genetics in order to identify how genetic factors contribute to phenotypic differences. This field has seen extensive research, with numerous methods proposed for predicting phenotypes. Nevertheless, the intricate relationship between genotypes and complex phenotypes, including common diseases, has resulted in an ongoing challenge to accurately decipher the genetic contribution. (2) Results: In this study, we propose a novel feature selection framework for phenotype prediction utilizing a genetic algorithm (FSF-GA) that effectively reduces the feature space to identify genotypes contributing to phenotype prediction. We provide a comprehensive vignette of our method and conduct extensive experiments using a widely used yeast dataset. (3) Conclusions: Our experimental results show that our proposed FSF-GA method delivers comparable phenotype prediction performance as compared to baseline methods, while providing features selected for predicting phenotypes. These selected feature sets can be used to interpret the underlying genetic architecture that contributes to phenotypic variation.

Data Science and Plant Metabolomics

The study of plant metabolism is one of the most complex tasks, mainly due to the huge amount and structural diversity of metabolites, as well as the fact that they react to changes in the environment and ultimately influence each other. Metabolic profiling is most often carried out using tools that include mass spectrometry (MS), which is one of the most powerful analytical methods. All this means that even when analyzing a single sample, we can obtain thousands of data. Data science has the potential to revolutionize our understanding of plant metabolism. This review demonstrates that machine learning, network analysis, and statistical modeling are some techniques being used to analyze large quantities of complex data that provide insights into plant development, growth, and how they interact with their environment. These findings could be key to improving crop yields, developing new forms of plant biotechnology, and understanding the relationship between plants and microbes. It is also necessary to consider the constraints that come with data science such as quality and availability of data, model complexity, and the need for deep knowledge of the subject in order to achieve reliable outcomes.

Simulated annealing aided genetic algorithm for gene selection from microarray data

In recent times, microarray gene expression datasets have gained significant popularity due to their usefulness to identify different types of cancer directly through bio-markers. These datasets possess a high gene-to-sample ratio and high dimensionality, with only a few genes functioning as bio-markers. Consequently, a significant amount of data is redundant, and it is essential to filter out important genes carefully. In this paper, we propose the Simulated Annealing aided Genetic Algorithm (SAGA), a meta-heuristic approach to identify informative genes from high-dimensional datasets. SAGA utilizes a two-way mutation-based Simulated Annealing (SA) as well as Genetic Algorithm (GA) to ensure a good trade-off between exploitation and exploration of the search space, respectively. The naive version of GA often gets stuck in a local optimum and depends on the initial population, leading to premature convergence. To address this, we have blended a clustering-based population generation with SA to distribute the initial population of GA over the entire feature space. To further enhance the performance, we reduce the initial search space by a score-based filter approach called the Mutually Informed Correlation Coefficient (MICC). The proposed method is evaluated on 6 microarray and 6 omics datasets. Comparison of SAGA with contemporary algorithms has shown that SAGA performs much better than its peers. Our code is available at https://github.com/shyammarjit/SAGA.

Genomic prediction in plants: opportunities for ensemble machine learning based approaches

Background: Many studies have demonstrated the utility of machine learning (ML) methods for genomic prediction (GP) of various plant traits, but a clear rationale for choosing ML over conventionally used, often simpler parametric methods, is still lacking. Predictive performance of GP models might depend on a plethora of factors including sample size, number of markers, population structure and genetic architecture. Methods: Here, we investigate which problem and dataset characteristics are related to good performance of ML methods for genomic prediction. We compare the predictive performance of two frequently used ensemble ML methods (Random Forest and Extreme Gradient Boosting) with parametric methods including genomic best linear unbiased prediction (GBLUP), reproducing kernel Hilbert space regression (RKHS), BayesA and BayesB. To explore problem characteristics, we use simulated and real plant traits under different genetic complexity levels determined by the number of Quantitative Trait Loci (QTLs), heritability (h2 and h2e), population structure and linkage disequilibrium between causal nucleotides and other SNPs. Results: Decision tree based ensemble ML methods are a better choice for nonlinear phenotypes and are comparable to Bayesian methods for linear phenotypes in the case of large effect Quantitative Trait Nucleotides (QTNs). Furthermore, we find that ML methods are susceptible to confounding due to population structure but less sensitive to low linkage disequilibrium than linear parametric methods. Conclusions: Overall, this provides insights into the role of ML in GP as well as guidelines for practitioners.

M6A-GSMS: Computational identification of N6-methyladenosine sites with GBDT and stacking learning in multiple species

N⁶-methyladenosine (m⁶A) is one of the most abundant forms of RNA methylation modifications currently known. It involves a wide range of biological processes, including degradation, stability, alternative splicing, etc. Therefore, the development of convenient and efficient m⁶A prediction technologies are urgent. In this work, a novel predictor based on GBDT and stacking learning is developed to identify m⁶A sites, which is called M6A-GSMS. To achieve accurate prediction, we explore RNA sequence information from four aspects: correlation, structure, physicochemical properties and pseudo ribonucleic acid composition. After using the GBDT algorithm for feature selection, a stacking model is constructed by combining seven basic classifiers. Compared with other state-of-the-art methods, the results show that M6A-GSMS can obtain excellent performance for identifying the m⁶A sites. The prediction accuracy of A.thaliana, D.melanogaster, M.musculus, S.cerevisiae and Human reaches 88.4%, 60.8%, 80.5%, 92.4% and 61.8%, respectively. This method provides an effective prediction for the investigation of m⁶A sites. In addition, all the datasets and codes are currently available at https://github.com/Wang-Jinyue/M6A-GSMS.Communicated by Ramaswamy H. Sarma.

Feature Selection Method Using Multi-Agent Reinforcement Learning Based on Guide Agents

In this study, we propose a method to automatically find features from a dataset that are effective for classification or prediction, using a new method called multi-agent reinforcement learning and a guide agent. Each feature of the dataset has one of the main and guide agents, and these agents decide whether to select a feature. Main agents select the optimal features, and guide agents present the criteria for judging the main agents' actions. After obtaining the main and guide rewards for the features selected by the agents, the main agent that behaves differently from the guide agent updates their Q-values by calculating the learning reward delivered to the main agents. The behavior comparison helps the main agent decide whether its own behavior is correct, without using other algorithms. After performing this process for each episode, the features are finally selected. The feature selection method proposed in this study uses multiple agents, reducing the number of actions each agent can perform and finding optimal features effectively and quickly. Finally, comparative experimental results on multiple datasets show that the proposed method can select effective features for classification and increase classification accuracy.

The influence of genetic structure on phenotypic diversity in the Australian mango (Mangifera indica) gene pool

Genomic selection is a promising breeding technique for tree crops to accelerate the development of new cultivars. However, factors such as genetic structure can create spurious associations between genotype and phenotype due to the shared history between populations with different trait values. Genetic structure can therefore reduce the accuracy of the genotype to phenotype map, a fundamental requirement of genomic selection models. Here, we employed 272 single nucleotide polymorphisms from 208 Mangifera indica accessions to explore whether the genetic structure of the Australian mango gene pool explained variation in trunk circumference, fruit blush colour and intensity. Multiple population genetic analyses indicate the presence of four genetic clusters and show that the most genetically differentiated cluster contains accessions imported from Southeast Asia (mainly those from Thailand). We find that genetic structure was strongly associated with three traits: trunk circumference, fruit blush colour and intensity in M. indica. This suggests that the history of these accessions could drive spurious associations between loci and key mango phenotypes in the Australian mango gene pool. Incorporating such genetic structure in associations between genotype and phenotype can improve the accuracy of genomic selection, which can assist the future development of new cultivars.

Do open citations give insights on the qualitative peer-review evaluation in research assessments? An analysis of the Italian National Scientific Qualification

In the past, several works have investigated ways for combining quantitative and qualitative methods in research assessment exercises. Indeed, the Italian National Scientific Qualification (NSQ), i.e. the national assessment exercise which aims at deciding whether a scholar can apply to professorial academic positions as Associate Professor and Full Professor, adopts a quantitative and qualitative evaluation process: it makes use of bibliometrics followed by a peer-review process of candidates’ CVs. The NSQ divides academic disciplines into two categories, i.e. citation-based disciplines (CDs) and non-citation-based disciplines (NDs), a division that affects the metrics used for assessing the candidates of that discipline in the first part of the process, which is based on bibliometrics. In this work, we aim at exploring whether citation-based metrics, calculated only considering open bibliographic and citation data, can support the human peer-review of NDs and yield insights on how it is conducted. To understand if and what citation-based (and, possibly, other) metrics provide relevant information, we created a series of machine learning models to replicate the decisions of the NSQ committees. As one of the main outcomes of our study, we noticed that the strength of the citational relationship between the candidate and the commission in charge of assessing his/her CV seems to play a role in the peer-review phase of the NSQ of NDs.

Data-driven learning to identify biomarkers in bipolar disorder

BACKGROUND AND OBJECTIVE

Bipolar disorder (BD) is one of the primary causes of disability globally and can be easily misdiagnosed as schizophrenia or major depression due to their similar symptoms. Hence, it is of great significance to explore the pathogenesis of BD. Statistical analysis is currently the most common method for exploring the neuropathological mechanisms of psychiatric disorders. However, this method only considers the relationship between groups and does not reflect the individual-level diagnosis. Therefore, we developed machine learning algorithms to measure pathological brain changes in psychiatric disorders.

METHODS

An autoencoder and a feature selection method are proposed to identify the abnormal structural patterns of BD in this study. The autoencoder was constructed using structural imaging data from 1113 healthy controls, which aims to define the normal range of anatomical deviations to distinguish healthy individuals from BD patients. The biomarkers of BD were identified by the reconstruction errors in each brain region. The proposed feature selection (FS)-select framework aimed to determine the optimal FS method and identify the most reproducible feature associated with BD.

RESULTS

We found that the left orbital region of the middle frontal gyrus had the greatest difference between healthy controls and BD patients using a trained autoencoder. The most reproducible feature was the left orbital region of the middle frontal gyrus by FS-select framework when using the different cross-validation strategies.

CONCLUSIONS

A consistent result was obtained from the above two proposed methods wherein a significant difference between healthy controls and BD patients was identified in the left orbital region of the middle frontal gyrus.

A divide-and-conquer approach for genomic prediction in rubber tree using machine learning

Rubber tree (Hevea brasiliensis) is the main feedstock for commercial rubber; however, its long vegetative cycle has hindered the development of more productive varieties via breeding programs. With the availability of H. brasiliensis genomic data, several linkage maps with associated quantitative trait loci have been constructed and suggested as a tool for marker-assisted selection. Nonetheless, novel genomic strategies are still needed, and genomic selection (GS) may facilitate rubber tree breeding programs aimed at reducing the required cycles for performance assessment. Even though such a methodology has already been shown to be a promising tool for rubber tree breeding, increased model predictive capabilities and practical application are still needed. Here, we developed a novel machine learning-based approach for predicting rubber tree stem circumference based on molecular markers. Through a divide-and-conquer strategy, we propose a neural network prediction system with two stages: (1) subpopulation prediction and (2) phenotype estimation. This approach yielded higher accuracies than traditional statistical models in a single-environment scenario. By delivering large accuracy improvements, our methodology represents a powerful tool for use in Hevea GS strategies. Therefore, the incorporation of machine learning techniques into rubber tree GS represents an opportunity to build more robust models and optimize Hevea breeding programs.

Early Prediction of Diabetes Using an Ensemble of Machine Learning Models

Diabetes is one of the most rapidly spreading diseases in the world, resulting in an array of significant complications, including cardiovascular disease, kidney failure, diabetic retinopathy, and neuropathy, among others, which contribute to an increase in morbidity and mortality rate. If diabetes is diagnosed at an early stage, its severity and underlying risk factors can be significantly reduced. However, there is a shortage of labeled data and the occurrence of outliers or data missingness in clinical datasets that are reliable and effective for diabetes prediction, making it a challenging endeavor. Therefore, we introduce a newly labeled diabetes dataset from a South Asian nation (Bangladesh). In addition, we suggest an automated classification pipeline that includes a weighted ensemble of machine learning (ML) classifiers: Naive Bayes (NB), Random Forest (RF), Decision Tree (DT), XGBoost (XGB), and LightGBM (LGB). Grid search hyperparameter optimization is employed to tune the critical hyperparameters of these ML models. Furthermore, missing value imputation, feature selection, and K-fold cross-validation are included in the framework design. A statistical analysis of variance (ANOVA) test reveals that the performance of diabetes prediction significantly improves when the proposed weighted ensemble (DT + RF + XGB + LGB) is executed with the introduced preprocessing, with the highest accuracy of 0.735 and an area under the ROC curve (AUC) of 0.832. In conjunction with the suggested ensemble model, our statistical imputation and RF-based feature selection techniques produced the best results for early diabetes prediction. Moreover, the presented new dataset will contribute to developing and implementing robust ML models for diabetes prediction utilizing population-level data.

Single nucleotide polymorphism marker combinations for classifying Yeonsan Ogye chicken using a machine learning approach

Genetic analysis has great potential as a tool to differentiate between different species and breeds of livestock. In this study, the optimal combinations of single nucleotide polymorphism (SNP) markers for discriminating the Yeonsan Ogye chicken (Gallus gallus domesticus) breed were identified using high-density 600K SNP array data. In 3,904 individuals from 198 chicken breeds, SNP markers specific to the target population were discovered through a case-control genome-wide association study (GWAS) and filtered out based on the linkage disequilibrium blocks. Significant SNP markers were selected by feature selection applying two machine learning algorithms: Random Forest (RF) and AdaBoost (AB). Using a machine learning approach, the 38 (RF) and 43 (AB) optimal SNP marker combinations for the Yeonsan Ogye chicken population demonstrated 100% accuracy. Hence, the GWAS and machine learning models used in this study can be efficiently utilized to identify the optimal combination of markers for discriminating target populations using multiple SNP markers.

Multichannel Acoustic Spectroscopy of the Human Body for Inviolable Biometric Authentication

Specific features of the human body, such as fingerprint, iris, and face, are extensively used in biometric authentication. Conversely, the internal structure and material features of the body have not been explored extensively in biometrics. Bioacoustics technology is suitable for extracting information about the internal structure and biological and material characteristics of the human body. Herein, we report a biometric authentication method that enables multichannel bioacoustic signal acquisition with a systematic approach to study the effects of selectively distilled frequency features, increasing the number of sensing channels with respect to multiple fingers. The accuracy of identity recognition according to the number of sensing channels and the number of selectively chosen frequency features was evaluated using exhaustive combination searches and forward-feature selection. The technique was applied to test the accuracy of machine learning classification using 5,232 datasets from 54 subjects. By optimizing the scanning frequency and sensing channels, our method achieved an accuracy of 99.62%, which is comparable to existing biometric methods. Overall, the proposed biometric method not only provides an unbreakable, inviolable biometric but also can be applied anywhere in the body and can substantially broaden the use of biometrics by enabling continuous identity recognition on various body parts for biometric identity authentication.

Two-stage improved Grey Wolf optimization algorithm for feature selection on high-dimensional classification

In recent years, evolutionary algorithms have shown great advantages in the field of feature selection because of their simplicity and potential global search capability. However, most of the existing feature selection algorithms based on evolutionary computation are wrapper methods, which are computationally expensive, especially for high-dimensional biomedical data. To significantly reduce the computational cost, it is essential to study an effective evaluation method. In this paper, a two-stage improved gray wolf optimization (IGWO) algorithm for feature selection on high-dimensional data is proposed. In the first stage, a multilayer perceptron (MLP) network with group lasso regularization terms is first trained to construct an integer optimization problem using the proposed algorithm for pre-selection of features and optimization of the hidden layer structure. The dataset is compressed using the feature subset obtained in the first stage. In the second stage, a multilayer perceptron network with group lasso regularization terms is retrained using the compressed dataset, and the proposed algorithm is employed to construct the discrete optimization problem for feature selection. Meanwhile, a rapid evaluation strategy is constructed to mitigate the evaluation cost and improve the evaluation efficiency in the feature selection process. The effectiveness of the algorithm was analyzed on ten gene expression datasets. The experimental results show that the proposed algorithm not only removes almost more than 95.7% of the features in all datasets, but also has better classification accuracy on the test set. In addition, the advantages of the proposed algorithm in terms of time consumption, classification accuracy and feature subset size become more and more prominent as the dimensionality of the feature selection problem increases. This indicates that the proposed algorithm is particularly suitable for solving high-dimensional feature selection problems.

Genomic prediction in plants: opportunities for ensemble machine learning based approaches

Background: Many studies have demonstrated the utility of machine learning (ML) methods for genomic prediction (GP) of various plant traits, but a clear rationale for choosing ML over conventionally used, often simpler parametric methods, is still lacking. Predictive performance of GP models might depend on a plethora of factors including sample size, number of markers, population structure and genetic architecture. Methods: Here, we investigate which problem and dataset characteristics are related to good performance of ML methods for genomic prediction. We compare the predictive performance of two frequently used ensemble ML methods (Random Forest and Extreme Gradient Boosting) with parametric methods including genomic best linear unbiased prediction (GBLUP), reproducing kernel Hilbert space regression (RKHS), BayesA and BayesB. To explore problem characteristics, we use simulated and real plant traits under different genetic complexity levels determined by the number of Quantitative Trait Loci (QTLs), heritability (h2 and h2e), population structure and linkage disequilibrium between causal nucleotides and other SNPs. Results: Decision tree based ensemble ML methods are a better choice for nonlinear phenotypes and are comparable to Bayesian methods for linear phenotypes in the case of large effect Quantitative Trait Nucleotides (QTNs). Furthermore, we find that ML methods are susceptible to confounding due to population structure but less sensitive to low linkage disequilibrium than linear parametric methods. Conclusions: Overall, this provides insights into the role of ML in GP as well as guidelines for practitioners.

Radiomics: a quantitative imaging biomarker in precision oncology

Cancer treatment is heading towards precision medicine driven by genetic and biochemical markers. Various genetic and biochemical markers are utilized to render personalized treatment in cancer. In the last decade, noninvasive imaging biomarkers have also been developed to assist personalized decision support systems in oncology. The imaging biomarkers i.e., radiomics is being researched to develop specific digital phenotype of tumor in cancer. Radiomics is a process to extract high throughput data from medical images by using advanced mathematical and statistical algorithms. The radiomics process involves various steps i.e., image generation, segmentation of region of interest (e.g. a tumor), image preprocessing, radiomic feature extraction, feature analysis and selection and finally prediction model development. Radiomics process explores the heterogeneity, irregularity and size parameters of the tumor to calculate thousands of advanced features. Our study investigates the role of radiomics in precision oncology. Radiomics research has witnessed a rapid growth in the last decade with several studies published that show the potential of radiomics in diagnosis and treatment outcome prediction in oncology. Several radiomics based prediction models have been developed and reported in the literature to predict various prediction endpoints i.e., overall survival, progression-free survival and recurrence in various cancer i.e., brain tumor, head and neck cancer, lung cancer and several other cancer types. Radiomics based digital phenotypes have shown promising results in diagnosis and treatment outcome prediction in oncology. In the coming years, radiomics is going to play a significant role in precision oncology.

A2BCF: An Automated ABC-Based Feature Selection Algorithm for Classification Models in an Education Application

Feature selection is an essential step of preprocessing in Machine Learning (ML) algorithms that can significantly impact the performance of ML models. It is considered one of the most crucial phases of automated ML (AutoML). Feature selection aims to find the optimal subset of features and remove the noninformative features from the dataset. Feature selection also reduces the computational time and makes the data more understandable to the learning model. There are various heuristic search strategies to address combinatorial optimization challenges. This paper develops an Automated Artificial Bee Colony-based algorithm for Feature Selection (A2BCF) to solve a classification problem. The application domain evaluating our proposed algorithm is education science, which solves a binary classification problem, namely, undergraduate student success. The modifications made to the original Artificial Bee Colony algorithm make the algorithm a well-performed approach.

Alterations in common marmoset gut microbiome associated with duodenal strictures

Chronic gastrointestinal (GI) diseases are the most common diseases in captive common marmosets (Callithrix jacchus). Despite standardized housing, diet and husbandry, a recently described gastrointestinal syndrome characterized by duodenal ulcers and strictures was observed in a subset of marmosets sourced from the New England Primate Research Center. As changes in the gut microbiome have been associated with GI diseases, the gut microbiome of 52 healthy, non-stricture marmosets (153 samples) were compared to the gut microbiome of 21 captive marmosets diagnosed with a duodenal ulcer/stricture (57 samples). No significant changes were observed using alpha diversity metrics, and while the community structure was significantly different when comparing beta diversity between healthy and stricture cases, the results were inconclusive due to differences observed in the dispersion of both datasets. Differences in the abundance of individual taxa using ANCOM, as stricture-associated dysbiosis was characterized by Anaerobiospirillum loss and Clostridium perfringens increases. To identify microbial and serum biomarkers that could help classify stricture cases, we developed models using machine learning algorithms (random forest, classification and regression trees, support vector machines and k-nearest neighbors) to classify microbiome, serum chemistry or complete blood count (CBC) data. Random forest (RF) models were the most accurate models and correctly classified strictures using either 9 ASVs (amplicon sequence variants), 4 serum chemistry tests or 6 CBC tests. Based on the RF model and ANCOM results, C. perfringens was identified as a potential causative agent associated with the development of strictures. Clostridium perfringens was also isolated by microbiological culture in 4 of 9 duodenum samples from marmosets with histologically confirmed strictures. Due to the enrichment of C. perfringens in situ, we analyzed frozen duodenal tissues using both 16S microbiome profiling and RNAseq. Microbiome analysis of the duodenal tissues of 29 marmosets from the MIT colony confirmed an increased abundance of Clostridium in stricture cases. Comparison of the duodenal gene expression from stricture and non-stricture marmosets found enrichment of genes associated with intestinal absorption, and lipid metabolism, localization, and transport in stricture cases. Using machine learning, we identified increased abundance of C. perfringens, as a potential causative agent of GI disease and intestinal strictures in marmosets.

Alternate fluency in Parkinson’s disease: A machine learning analysis

Objective

The aim of the present study was to investigate whether patients with Parkinson’s Disease (PD) had changes in their level of performance in extra-dimensional shifting by implementing a novel analysis method, utilizing the new alternate phonemic/semantic fluency test.

Method

We used machine learning (ML) in order to develop high accuracy classification between PD patients with high and low scores in the alternate fluency test.

Results

The models developed resulted to be accurate in such classification in a range between 80% and 90%. The predictor which demonstrated maximum efficiency in classifying the participants as low or high performers was the semantic fluency test. The optimal cut-off of a decision rule based on this test yielded an accuracy of 86.96%. Following the removal of the semantic fluency test from the system, the parameter which best contributed to the classification was the phonemic fluency test. The best cut-offs were identified and the decision rule yielded an overall accuracy of 80.43%. Lastly, in order to evaluate the classification accuracy based on the shifting index, the best cut-offs based on an optimal single rule yielded an overall accuracy of 83.69%.

Conclusion

We found that ML analysis of semantic and phonemic verbal fluency may be used to identify simple rules with high accuracy and good out of sample generalization, allowing the detection of executive deficits in patients with PD.

Identifying COVID-19 by using spectral analysis of cough recordings: a distinctive classification study

Sound signals from the respiratory system are largely taken as tokens of human health. Early diagnosis of respiratory tract diseases is of great importance because, if delayed, it exerts irreversible effects on human health. The Coronavirus pandemic, which is deeply shaking the world, has revealed the importance of this diagnosis even more. During the pandemic, it has become the focus of researchers to differentiate symptoms from similar diseases such as influenza. Among these symptoms, the difference in cough sound played a distinctive role in research. Clinical data collected under the supervision of doctors in a reliable environment were used as the dataset consisting of 16 subjects suspected of COVID-19 with a specific patient demographic. Using the polymerase chain reaction test, the suspected subjects were divided into two groups as negative and positive. The negative and positive labels represent the patients with non-COVID and with a COVID-19 cough, respectively. Using the 3D plot or waterfall representation of the signal frequency spectrum, the salient features of the cough data are revealed. In this way, COVID-19 can be differentiated from other coughs by applying effective feature extraction and classification techniques. Power spectral density based on short-time Fourier transform and mel-frequency cepstral coefficients (MFCC) were chosen as the efficient feature extraction method. From among the classification techniques, the support vector machine (SVM) algorithm was applied to the processed signals in order to identify and classify COVID-19 cough. In terms of results evaluation, the cough of subjects with COVID-19 was detected with 95.86% classification accuracy thanks to the radial basis function (RBF) kernel function of SVM and the MFCC method. The diagnosis of COVID-19 coughs was performed with 98.6% and 91.7% sensitivity and specificity, respectively.

Sleep Quality Modulates the Association between Dynamic Functional Network Connectivity and Cognitive Function in Healthy Older Adults

Aging is associated with changes in sleep, brain activity, and cognitive function, as well as the association among these factors; however, the precise nature of these changes has not been elucidated. This study systematically investigated the modulatory effect of sleep on the relationship between brain functional network connectivity (FNC) and cognitive function in older adults. In total, 107 community-dwelling healthy older adults were recruited and assigned into poor sleep and good sleep groups based on the Pittsburgh Sleep Quality Index. The static functional network connectivity (sFNC), the temporal variability of dynamic FNC (dFNC) from variance (dFNC-var), and the dFNC from clustering state (dFNC-state) were calculated. Corresponding cognition-predictive models were constructed for each sleep group. dFNC but not sFNC, was able to significantly predict the cognitive function in older adults. Specifically, sleep played a modulatory role in the association between dFNC and cognitive function, with sleep-specific variations at both microscopic (i.e., specific edges) and macroscopic levels (i.e., specific states) of dFNC.

QSAR analysis of pyrimidine derivatives as VEGFR-2 receptor inhibitors to inhibit cancer using multiple linear regression and artificial neural network

BACKGROUND AND PURPOSE

In this study, the pharmacological activity of 33 compounds of furopyrimidine and thienopyrimidine as vascular endothelial growth factor receptor 2 (VEGFR-2) inhibitors to inhibit cancer was investigated. The most important angiogenesis inducer is VEGF endothelial growth factor, which exerts its activity by binding to two tyrosine kinase receptors called VEGFR-1 and VEGFR-2. Due to the critical role of VEGF in the pathological angiogenesis of this molecule, it is a valuable therapeutic target for anti-angiogenesis therapies.

EXPERIMENTAL APPROACH

After calculating descriptors using SPSS software and stepwise selection method, 5 descriptors were used for modeling in multiple linear regression (MLR) and artificial neural network (ANN). The calibration series and the test series in this study included 26 and 7 combinations, respectively.

FINDINGS/RESULTS

The performance evaluation of models was determined by the R2, RMSE, and Q2 statistic parameters. The R2 values of MLR and ANN models were 0.889 and 0.998, respectively. Also, the value of RMSE in the ANN model was lower and its Q2 value was higher than the MLR model.

CONCLUSION AND IMPLICATIONS

The results were evaluated by different statistical methods and it was concluded that the nonlinear neural network method is powerful to predict the pharmacological activity of similar compounds, and because of the complex and nonlinear relationships, the MLR was not capable of establishing a good model with high predictive power.

Machine learning in the prediction of depression treatment outcomes: a systematic review and meta-analysis

BACKGROUND

Multiple treatments are effective for major depressive disorder (MDD), but the outcomes of each treatment vary broadly among individuals. Accurate prediction of outcomes is needed to help select a treatment that is likely to work for a given person. We aim to examine the performance of machine learning methods in delivering replicable predictions of treatment outcomes.

METHODS

Of 7732 non-duplicate records identified through literature search, we retained 59 eligible reports and extracted data on sample, treatment, predictors, machine learning method, and treatment outcome prediction. A minimum sample size of 100 and an adequate validation method were used to identify adequate-quality studies. The effects of study features on prediction accuracy were tested with mixed-effects models. Fifty-four of the studies provided accuracy estimates or other estimates that allowed calculation of balanced accuracy of predicting outcomes of treatment.

RESULTS

Eight adequate-quality studies reported a mean accuracy of 0.63 [95% confidence interval (CI) 0.56-0.71], which was significantly lower than a mean accuracy of 0.75 (95% CI 0.72-0.78) in the other 46 studies. Among the adequate-quality studies, accuracies were higher when predicting treatment resistance (0.69) and lower when predicting remission (0.60) or response (0.56). The choice of machine learning method, feature selection, and the ratio of features to individuals were not associated with reported accuracy.

CONCLUSIONS

The negative relationship between study quality and prediction accuracy, combined with a lack of independent replication, invites caution when evaluating the potential of machine learning applications for personalizing the treatment of depression.

Application of Machine Learning for eutrophication analysis and algal bloom prediction in an urban river: A 10-year study of the Han River, South Korea

The increasing release of nutrients to aquatic environments has led to great concern regarding eutrophication and the risk of unwanted algal blooms. Based on observational data of 20 water quality parameters measured on a monthly basis at 40 stations from 2011 to 2020, this study applied different Machine Learning (ML) algorithms to suggest the best option for algal bloom prediction in the Han River, a large river in South Korea. Eight different ML algorithms were categorized into several groups of statistical learning, regression family, and deep learning, and were then compared for their suitability to predict the chlorophyll-derived trophic index (TSI-Chla). ML algorithms helped identify the most important water quality parameters contributing to algal bloom prediction. The ML results confirmed that eutrophication and algal proliferation were governed by the complex interplay between nutrients (nitrogen and phosphorus), organic contaminants, and environmental factors. Of the models tested, the adaptive neuro-fuzzy inference system (ANFIS) exhibited the best performance owing to its consistent and outperforming prediction both quantitatively (i.e., via regression) and qualitatively (i.e., via classification), which was evidenced by the lowest value of mean absolute error (MAE) of 0.09, and the highest F1-score, Recall and Precision of 0.97, 0.98 and 0.96, respectively. In a further step, a representative web application was constructed to assist common users to predict the trophic status of the Han River. This study demonstrated that ML techniques are not only promising for highly accurate water quality modeling of urban rivers, but also reduce time and labor intensity for experiments, which decreases the number of monitored water quality parameters, providing further insights into the driving factors of water quality deterioration. They ultimately help devise proactive strategies for sustainable water management.

Differentiation of Cystic Fibrosis-Related Pathogens by Volatile Organic Compound Analysis with Secondary Electrospray Ionization Mass Spectrometry

Identifying and differentiating bacteria based on their emitted volatile organic compounds (VOCs) opens vast opportunities for rapid diagnostics. Secondary electrospray ionization high-resolution mass spectrometry (SESI-HRMS) is an ideal technique for VOC-biomarker discovery because of its speed, sensitivity towards polar molecules and compound characterization possibilities. Here, an in vitro SESI-HRMS workflow to find biomarkers for cystic fibrosis (CF)-related pathogens P. aeruginosa, S. pneumoniae, S. aureus, H. influenzae, E. coli and S. maltophilia is described. From 180 headspace samples, the six pathogens are distinguishable in the first three principal components and predictive analysis with a support vector machine algorithm using leave-one-out cross-validation exhibited perfect accuracy scores for the differentiation between the groups. Additionally, 94 distinctive features were found by recursive feature elimination and further characterized by SESI-MS/MS, which yielded 33 putatively identified biomarkers. In conclusion, the six pathogens can be distinguished in vitro based on their VOC profiles as well as the herein reported putative biomarkers. In the future, these putative biomarkers might be helpful for pathogen detection in vivo based on breath samples from patients with CF.

From genetics to systems biology of stress-related mental disorders

Many individuals will be exposed to some form of traumatic stress in their lifetime which, in turn, increases the likelihood of developing stress-related disorders such as post-traumatic stress disorder (PTSD), major depressive disorder (MDD) and anxiety disorders (ANX). The development of these disorders is also influenced by genetics and have heritability estimates ranging between ∼30 and 70%. In this review, we provide an overview of the findings of genome-wide association studies for PTSD, depression and ANX, and we observe a clear genetic overlap between these three diagnostic categories. We go on to highlight the results from transcriptomic and epigenomic studies, and, given the multifactorial nature of stress-related disorders, we provide an overview of the gene-environment studies that have been conducted to date. Finally, we discuss systems biology approaches that are now seeing wider utility in determining a more holistic view of these complex disorders.

A primer on machine learning techniques for genomic applications

High throughput sequencing technologies have enabled the study of complex biological aspects at single nucleotide resolution, opening the big data era. The analysis of large volumes of heterogeneous "omic" data, however, requires novel and efficient computational algorithms based on the paradigm of Artificial Intelligence. In the present review, we introduce and describe the most common machine learning methodologies, and lately deep learning, applied to a variety of genomics tasks, trying to emphasize capabilities, strengths and limitations through a simple and intuitive language. We highlight the power of the machine learning approach in handling big data by means of a real life example, and underline how described methods could be relevant in all cases in which large amounts of multimodal genomic data are available.

Two-stage improved Grey Wolf optimization algorithm for feature selection on high-dimensional classification

In recent years, evolutionary algorithms have shown great advantages in the field of feature selection because of their simplicity and potential global search capability. However, most of the existing feature selection algorithms based on evolutionary computation are wrapper methods, which are computationally expensive, especially for high-dimensional biomedical data. To significantly reduce the computational cost, it is essential to study an effective evaluation method. In this paper, a two-stage improved gray wolf optimization (IGWO) algorithm for feature selection on high-dimensional data is proposed. In the first stage, a multilayer perceptron (MLP) network with group lasso regularization terms is first trained to construct an integer optimization problem using the proposed algorithm for pre-selection of features and optimization of the hidden layer structure. The dataset is compressed using the feature subset obtained in the first stage. In the second stage, a multilayer perceptron network with group lasso regularization terms is retrained using the compressed dataset, and the proposed algorithm is employed to construct the discrete optimization problem for feature selection. Meanwhile, a rapid evaluation strategy is constructed to mitigate the evaluation cost and improve the evaluation efficiency in the feature selection process. The effectiveness of the algorithm was analyzed on ten gene expression datasets. The experimental results show that the proposed algorithm not only removes almost more than 95.7% of the features in all datasets, but also has better classification accuracy on the test set. In addition, the advantages of the proposed algorithm in terms of time consumption, classification accuracy and feature subset size become more and more prominent as the dimensionality of the feature selection problem increases. This indicates that the proposed algorithm is particularly suitable for solving high-dimensional feature selection problems.

Perspective: Big Data and Machine Learning Could Help Advance Nutritional Epidemiology

The field of nutritional epidemiology faces challenges posed by measurement error, diet as a complex exposure, and residual confounding. The objective of this perspective article is to highlight how developments in big data and machine learning can help address these challenges. New methods of collecting 24-h dietary recalls and recording diet could enable larger samples and more repeated measures to increase statistical power and measurement precision. In addition, use of machine learning to automatically classify pictures of food could become a useful complimentary method to help improve precision and validity of dietary measurements. Diet is complex due to thousands of different foods that are consumed in varying proportions, fluctuating quantities over time, and differing combinations. Current dietary pattern methods may not integrate sufficient dietary variation, and most traditional modeling approaches have limited incorporation of interactions and nonlinearity. Machine learning could help better model diet as a complex exposure with nonadditive and nonlinear associations. Last, novel big data sources could help avoid unmeasured confounding by offering more covariates, including both omics and features derived from unstructured data with machine learning methods. These opportunities notwithstanding, application of big data and machine learning must be approached cautiously to ensure quality of dietary measurements, avoid overfitting, and confirm accurate interpretations. Greater use of machine learning and big data would also require substantial investments in training, collaborations, and computing infrastructure. Overall, we propose that judicious application of big data and machine learning in nutrition science could offer new means of dietary measurement, more tools to model the complexity of diet and its relations with diseases, and additional potential ways of addressing confounding.

Weight Feedback-Based Harmonic MDG-Ensemble Model for Prediction of Traffic Accident Severity

Traffic accidents are emerging as a serious social problem in modern society but if the severity of an accident is quickly grasped, countermeasures can be organized efficiently. To solve this problem, the method proposed in this paper derives the MDG (Mean Decrease Gini) coefficient between variables to assess the severity of traffic accidents. Single models are designed to use coefficient, independent variables to determine and predict accident severity. The generated single models are fused using a weighted-voting-based bagging method ensemble to consider various characteristics and avoid overfitting. The variables used for predicting accidents are classified as dependent or independent and the variables that affect the severity of traffic accidents are predicted using the characteristics of causal relationships. Independent variables are classified as categorical and numerical variables. For this reason, a problem arises when the variation among dependent variables is imbalanced. Therefore, a harmonic average is applied to the weights to maintain the variables’ balance and determine the average rate of change. Through this, it is possible to establish objective criteria for determining the severity of traffic accidents, thereby improving reliability.

Reproducible Evaluation of Diffusion MRI Features for Automatic Classification of Patients with Alzheimer's Disease

Diffusion MRI is the modality of choice to study alterations of white matter. In past years, various works have used diffusion MRI for automatic classification of Alzheimer's disease. However, classification performance obtained with different approaches is difficult to compare because of variations in components such as input data, participant selection, image preprocessing, feature extraction, feature rescaling (FR), feature selection (FS) and cross-validation (CV) procedures. Moreover, these studies are also difficult to reproduce because these different components are not readily available. In a previous work (Samper-González et al. 2018), we propose an open-source framework for the reproducible evaluation of AD classification from T1-weighted (T1w) MRI and PET data. In the present paper, we first extend this framework to diffusion MRI data. Specifically, we add: conversion of diffusion MRI ADNI data into the BIDS standard and pipelines for diffusion MRI preprocessing and feature extraction. We then apply the framework to compare different components. First, FS has a positive impact on classification results: highest balanced accuracy (BA) improved from 0.76 to 0.82 for task CN vs AD. Secondly, voxel-wise features generally gives better performance than regional features. Fractional anisotropy (FA) and mean diffusivity (MD) provided comparable results for voxel-wise features. Moreover, we observe that the poor performance obtained in tasks involving MCI were potentially caused by the small data samples, rather than by the data imbalance. Furthermore, no extensive classification difference exists for different degree of smoothing and registration methods. Besides, we demonstrate that using non-nested validation of FS leads to unreliable and over-optimistic results: 5% up to 40% relative increase in BA. Lastly, with proper FR and FS, the performance of diffusion MRI features is comparable to that of T1w MRI. All the code of the framework and the experiments are publicly available: general-purpose tools have been integrated into the Clinica software package ( www.clinica.run ) and the paper-specific code is available at: https://github.com/aramis-lab/AD-ML .