Mining gene expression data of multiple sclerosis

The role of machine learning in developing non-magnetic resonance imaging based biomarkers for multiple sclerosis: a systematic review

BACKGROUND

Multiple sclerosis (MS) is a neurological condition whose symptoms, severity, and progression over time vary enormously among individuals. Ideally, each person living with MS should be provided with an accurate prognosis at the time of diagnosis, precision in initial and subsequent treatment decisions, and improved timeliness in detecting the need to reassess treatment regimens. To manage these three components, discovering an accurate, objective measure of overall disease severity is essential. Machine learning (ML) algorithms can contribute to finding such a clinically useful biomarker of MS through their ability to search and analyze datasets about potential biomarkers at scale. Our aim was to conduct a systematic review to determine how, and in what way, ML has been applied to the study of MS biomarkers on data from sources other than magnetic resonance imaging.

METHODS

Systematic searches through eight databases were conducted for literature published in 2014-2020 on MS and specified ML algorithms.

RESULTS

Of the 1, 052 returned papers, 66 met the inclusion criteria. All included papers addressed developing classifiers for MS identification or measuring its progression, typically, using hold-out evaluation on subsets of fewer than 200 participants with MS. These classifiers focused on biomarkers of MS, ranging from those derived from omics and phenotypical data (34.5% clinical, 33.3% biological, 23.0% physiological, and 9.2% drug response). Algorithmic choices were dependent on both the amount of data available for supervised ML (91.5%; 49.2% classification and 42.3% regression) and the requirement to be able to justify the resulting decision-making principles in healthcare settings. Therefore, algorithms based on decision trees and support vector machines were commonly used, and the maximum average performance of 89.9% AUC was found in random forests comparing with other ML algorithms.

CONCLUSIONS

ML is applicable to determining how candidate biomarkers perform in the assessment of disease severity. However, applying ML research to develop decision aids to help clinicians optimize treatment strategies and analyze treatment responses in individual patients calls for creating appropriate data resources and shared experimental protocols. They should target proceeding from segregated classification of signals or natural language to both holistic analyses across data modalities and clinically-meaningful differentiation of disease.

Multiple sclerosis diagnosis and phenotype identification by multivariate classification of in vivo frontal cortex metabolite profiles

Multiple sclerosis (MS) is a heterogeneous autoimmune disease for which diagnosis continues to rely on subjective clinical judgment over a battery of tests. Proton magnetic resonance spectroscopy (¹H MRS) enables the noninvasive in vivo detection of multiple small-molecule metabolites and is therefore in principle a promising means of gathering information sufficient for multiple sclerosis diagnosis and subtype classification. Here we show that supervised classification using ¹H-MRS-visible normal-appearing frontal cortex small-molecule metabolites alone can indeed differentiate individuals with progressive MS from control (held-out validation sensitivity 79% and specificity 68%), as well as between relapsing and progressive MS phenotypes (held-out validation sensitivity 84% and specificity 74%). Post hoc assessment demonstrated the disproportionate contributions of glutamate and glutamine to identifying MS status and phenotype, respectively. Our finding establishes ¹H MRS as a viable means of characterizing progressive multiple sclerosis disease status and paves the way for continued refinement of this method as an auxiliary or mainstay of multiple sclerosis diagnostics.

An Enhanced Multiple Sclerosis Disease Diagnosis via an Ensemble Approach

Multiple Sclerosis (MS) is a disease attacking the central nervous system. According to MS Atlas’s most recent statistics, there are more than 2.8 million people worldwide diagnosed with MS. Recently, studies started to explore machine learning techniques to predict MS using various data. The objective of this paper is to develop an ensemble approach for diagnosis of MS using gene expression profiles, while handling the class imbalance problem associated with the data. A hierarchical ensemble approach employing voting and boosting techniques is proposed. This approach adopts a heterogeneous voting approach using two base learners, random forest and support vector machine. Experiments show that our approach outperforms state-of-the-art methods, with the highest recorded accuracy being 92.81% and 93.5% with BoostFS and DEGs for feature selection, respectively. Conclusively, the proposed approach is able to efficiently diagnose MS using the gene expression profiles that are more relevant to the disease. The approach is not merely an ensemble classifier outperforming previous work; it also identifies differentially expressed genes between normal samples and patients with multiple sclerosis using a genome-wide expression microarray. The results obtained show that the proposed approach is an efficient diagnostic tool for MS.

Deep learning‐based microarray cancer classification and ensemble gene selection approach

Malignancies and diseases of various genetic origins can be diagnosed and classified with microarray data. There are many obstacles to overcome due to the large size of the gene and the small number of samples in the microarray. A combination strategy for gene expression in a variety of diseases is described in this paper, consisting of two steps: identifying the most effective genes via soft ensembling and classifying them with a novel deep neural network. The feature selection approach combines three strategies to select wrapper genes and rank them according to the k‐nearest neighbour algorithm, resulting in a very generalisable model with low error levels. Using soft ensembling, the most effective subsets of genes were identified from three microarray datasets of diffuse large cell lymphoma, leukaemia, and prostate cancer. A stacked deep neural network was used to classify all three datasets, achieving an average accuracy of 97.51%, 99.6%, and 96.34%, respectively. In addition, two previously unreported datasets from small, round blue cell tumors (SRBCTs)and multiple sclerosis‐related brain tissue lesions were examined to show the generalisability of the model method.

Novel Insight Into the Role of Immune Dysregulation in Amyotrophic Lateral Sclerosis Based on Bioinformatic Analysis

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disorder characterized by the progressive degeneration of motor neurons. The causative pathogenic mechanisms in ALS remain unclear, limiting the development of treatment strategies. Neuroinflammation and immune dysregulation were involved in the disease onset and progression of several neurodegenerative disorders, including ALS. In this study, we carried out a bioinformatic analysis using publicly available datasets from Gene Expression Omnibus (GEO) to investigate the role of immune cells and genes alterations in ALS. Single-sample gene set enrichment analysis revealed that the infiltration of multiple types of immune cells, including macrophages, type-1/17 T helper cells, and activated CD4 + /CD8 + T cells, was higher in ALS patients than in controls. Weighted gene correlation network analysis identified immune genes associated with ALS. The Gene Ontology analysis revealed that receptor and cytokine activities were the most highly enriched terms. Pathway analysis showed that these genes were enriched not only in immune-related pathways, such as cytokine-cytokine receptor interaction, but also in PI3K-AKT and MAPK signaling pathways. Nineteen immune-related genes (C3AR1, CCR1, CCR5, CD86, CYBB, FCGR2B, FCGR3A, HCK, ITGB2, PTPRC, TLR1, TLR2, TLR7, TLR8, TYROBP, VCAM1, CD14, CTSS, and FCER1G) were identified as hub genes based on least absolute shrinkage and selection operator analysis. This gene signature could differentiate ALS patients from non-neurological controls (p < 0.001) and predict disease occurrence (AUC = 0.829 in training set; AUC = 0.862 in test set). In conclusion, our study provides potential biomarkers of ALS for disease diagnosis and therapeutic monitoring.

Machine Learning Use for Prognostic Purposes in Multiple Sclerosis

The course of multiple sclerosis begins with a relapsing-remitting phase, which evolves into a secondarily progressive form over an extremely variable period, depending on many factors, each with a subtle influence. To date, no prognostic factors or risk score have been validated to predict disease course in single individuals. This is increasingly frustrating, since several treatments can prevent relapses and slow progression, even for a long time, although the possible adverse effects are relevant, in particular for the more effective drugs. An early prediction of disease course would allow differentiation of the treatment based on the expected aggressiveness of the disease, reserving high-impact therapies for patients at greater risk. To increase prognostic capacity, approaches based on machine learning (ML) algorithms are being attempted, given the failure of other approaches. Here we review recent studies that have used clinical data, alone or with other types of data, to derive prognostic models. Several algorithms that have been used and compared are described. Although no study has proposed a clinically usable model, knowledge is building up and in the future strong tools are likely to emerge.

Factors Associated with E-Cigarette Use in U.S. Young Adult Never Smokers of Conventional Cigarettes: A Machine Learning Approach

E-cigarette use is increasing among young adult never smokers of conventional cigarettes, but the awareness of the factors associated with e-cigarette use in this population is limited. The goal of this work was to use machine learning (ML) algorithms to determine the factors associated with current e-cigarette use among US young adult never cigarette smokers. Young adult (18-34 years) never cigarette smokers from the 2016 and 2017 Behavioral Risk Factor Surveillance System (BRFSS) who reported current or never e-cigarette use were used for the analysis (n = 79,539). Variables associated with current e-cigarette use were selected by two ML algorithms (Boruta and Least absolute shrinkage and selection operator (LASSO)). Odds ratios were calculated to determine the association between e-cigarette use and the variables selected by the ML algorithms, after adjusting for age, gender and race/ethnicity and incorporating the BRFSS complex design. The prevalence of e-cigarette use varied across states. Factors previously reported in the literature, such as age, race/ethnicity, alcohol use, depression, as well as novel factors associated with e-cigarette use, such as disabilities, obesity, history of diabetes and history of arthritis were identified. These results can be used to generate further hypotheses for research, increase public awareness and help provide targeted e-cigarette education.

Rps27a might act as a controller of microglia activation in triggering neurodegenerative diseases

Neurodegenerative diseases (NDDs) are increasing serious menaces to human health in the recent years. Despite exhibiting different clinical phenotypes and selective neuronal loss, there are certain common features in these disorders, suggesting the presence of commonly dysregulated pathways. Identifying causal genes and dysregulated pathways can be helpful in providing effective treatment in these diseases. Interestingly, in spite of the considerable researches on NDDs, to the best of our knowledge, no dysregulated genes and/or pathways were reported in common across all the major NDDs so far. In this study, for the first time, we have applied the three-way interaction model, as an approach to unravel sophisticated gene interactions, to trace switch genes and significant pathways that are involved in six major NDDs. Subsequently, a gene regulatory network was constructed to investigate the regulatory communication of statistically significant triplets. Finally, KEGG pathway enrichment analysis was applied to find possible common pathways. Because of the central role of neuroinflammation and immune system responses in both pathogenic and protective mechanisms in the NDDs, we focused on immune genes in this study. Our results suggest that "cytokine-cytokine receptor interaction" pathway is enriched in all of the studied NDDs, while "osteoclast differentiation" and "natural killer cell mediated cytotoxicity" pathways are enriched in five of the NDDs each. The results of this study indicate that three pathways that include "osteoclast differentiation", "natural killer cell mediated cytotoxicity" and "cytokine-cytokine receptor interaction" are common in five, five and six NDDs, respectively. Additionally, our analysis showed that Rps27a as a switch gene, together with the gene pair {Il-18, Cx3cl1} form a statistically significant and biologically relevant triplet in the major NDDs. More specifically, we suggested that Cx3cl1 might act as a potential upstream regulator of Il-18 in microglia activation, and in turn, might be controlled with Rps27a in triggering NDDs.

Improvement in prediction of antigenic epitopes using stacked generalisation: an ensemble approach

The major intent of peptide vaccine designs, immunodiagnosis and antibody productions is to accurately identify linear B-cell epitopes. The determination of epitopes through experimental analysis is highly expensive. Therefore, it is desirable to develop a reliable model with significant improvement in prediction models. In this study, a hybrid model has been designed by using stacked generalisation ensemble technique for prediction of linear B-cell epitopes. The goal of using stacked generalisation ensemble approach is to refine predictions of base classifiers and to get rid of the worse predictions. In this study, six machine learning models are fused to predict variable length epitopes (6-49 mers). The proposed ensemble model achieves 76.6% accuracy and average accuracy of repeated 10-fold cross-validation is 73.14%. The trained ensemble model has been tested on the benchmark dataset and compared with existing sequential B-cell epitope prediction techniques including APCpred, ABCpred, BCpred and [inline-formula removed].

Identification of Genes Discriminating Multiple Sclerosis Patients from Controls by Adapting a Pathway Analysis Method

The focus of analyzing data from microarray experiments has shifted from the identification of associated individual genes to that of associated biological pathways or gene sets. In bioinformatics, a feature selection algorithm is usually used to cope with the high dimensionality of microarray data. In addition to those algorithms that use the biological information contained within a gene set as a priori to facilitate the process of feature selection, various gene set analysis methods can be applied directly or modified readily for the purpose of feature selection. Significance analysis of microarray to gene-set reduction analysis (SAM-GSR) algorithm, a novel direction of gene set analysis, is one of such methods. Here, we explore the feature selection property of SAM-GSR and provide a modification to better achieve the goal of feature selection. In a multiple sclerosis (MS) microarray data application, both SAM-GSR and our modification of SAM-GSR perform well. Our results show that SAM-GSR can carry out feature selection indeed, and modified SAM-GSR outperforms SAM-GSR. Given pathway information is far from completeness, a statistical method capable of constructing biologically meaningful gene networks is of interest. Consequently, both SAM-GSR algorithms will be continuously revaluated in our future work, and thus better characterized.

Mining potential biomarkers associated with space flight in Caenorhabditis elegans experienced Shenzhou-8 mission with multiple feature selection techniques

To identify the potential biomarkers associated with space flight, a combined algorithm, which integrates the feature selection techniques, was used to deal with the microarray datasets of Caenorhabditis elegans obtained in the Shenzhou-8 mission. Compared with the ground control treatment, a total of 86 differentially expressed (DE) genes in responses to space synthetic environment or space radiation environment were identified by two filter methods. And then the top 30 ranking genes were selected by the random forest algorithm. Gene Ontology annotation and functional enrichment analyses showed that these genes were mainly associated with metabolism process. Furthermore, clustering analysis showed that 17 genes among these are positive, including 9 for space synthetic environment and 8 for space radiation environment only. These genes could be used as the biomarkers to reflect the space environment stresses. In addition, we also found that microgravity is the main stress factor to change the expression patterns of biomarkers for the short-duration spaceflight.

Integrated Analyses of Gene Expression Profiles Digs out Common Markers for Rheumatic Diseases

OBJECTIVE

Rheumatic diseases have some common symptoms. Extensive gene expression studies, accumulated thus far, have successfully identified signature molecules for each rheumatic disease, individually. However, whether there exist shared factors across rheumatic diseases has yet to be tested.

METHODS

We collected and utilized 6 public microarray datasets covering 4 types of representative rheumatic diseases including rheumatoid arthritis, systemic lupus erythematosus, ankylosing spondylitis, and osteoarthritis. Then we detected overlaps of differentially expressed genes across datasets and performed a meta-analysis aiming at identifying common differentially expressed genes that discriminate between pathological cases and normal controls. To further gain insights into the functions of the identified common differentially expressed genes, we conducted gene ontology enrichment analysis and protein-protein interaction analysis.

RESULTS

We identified a total of eight differentially expressed genes (TNFSF10, CX3CR1, LY96, TLR5, TXN, TIA1, PRKCH, PRF1), each associated with at least 3 of the 4 studied rheumatic diseases. Meta-analysis warranted the significance of the eight genes and highlighted the general significance of four genes (CX3CR1, LY96, TLR5, and PRF1). Protein-protein interaction and gene ontology enrichment analyses indicated that the eight genes interact with each other to exert functions related to immune response and immune regulation.

CONCLUSION

The findings support that there exist common factors underlying rheumatic diseases. For rheumatoid arthritis, systemic lupus erythematosus, ankylosing spondylitis and osteoarthritis diseases, those common factors include TNFSF10, CX3CR1, LY96, TLR5, TXN, TIA1, PRKCH, and PRF1. In-depth studies on these common factors may provide keys to understanding the pathogenesis and developing intervention strategies for rheumatic diseases.

Improved Variable Selection Algorithm Using a LASSO-Type Penalty, with an Application to Assessing Hepatitis B Infection Relevant Factors in Community Residents

OBJECTIVES

In epidemiological studies, it is important to identify independent associations between collective exposures and a health outcome. The current stepwise selection technique ignores stochastic errors and suffers from a lack of stability. The alternative LASSO-penalized regression model can be applied to detect significant predictors from a pool of candidate variables. However, this technique is prone to false positives and tends to create excessive biases. It remains challenging to develop robust variable selection methods and enhance predictability.

MATERIAL AND METHODS

Two improved algorithms denoted the two-stage hybrid and bootstrap ranking procedures, both using a LASSO-type penalty, were developed for epidemiological association analysis. The performance of the proposed procedures and other methods including conventional LASSO, Bolasso, stepwise and stability selection models were evaluated using intensive simulation. In addition, methods were compared by using an empirical analysis based on large-scale survey data of hepatitis B infection-relevant factors among Guangdong residents.

RESULTS

The proposed procedures produced comparable or less biased selection results when compared to conventional variable selection models. In total, the two newly proposed procedures were stable with respect to various scenarios of simulation, demonstrating a higher power and a lower false positive rate during variable selection than the compared methods. In empirical analysis, the proposed procedures yielding a sparse set of hepatitis B infection-relevant factors gave the best predictive performance and showed that the procedures were able to select a more stringent set of factors. The individual history of hepatitis B vaccination, family and individual history of hepatitis B infection were associated with hepatitis B infection in the studied residents according to the proposed procedures.

CONCLUSIONS

The newly proposed procedures improve the identification of significant variables and enable us to derive a new insight into epidemiological association analysis.

Multiple sclerosis patient-derived CSF induces transcriptional changes in proliferating oligodendrocyte progenitors

BACKGROUND

Cerebrospinal fluid (CSF) is in contact with brain parenchyma and ventricles, and its composition might influence the cellular physiology of oligodendrocyte progenitor cells (OPCs) thereby contributing to multiple sclerosis (MS) disease pathogenesis.

OBJECTIVE

To identify the transcriptional changes that distinguish the transcriptional response induced in proliferating rat OPCs upon exposure to CSF from primary progressive multiple sclerosis (PPMS) or relapsing remitting multiple sclerosis (RRMS) patients and other neurological controls.

METHODS

We performed gene microarray analysis of OPCs exposed to CSF from neurological controls, or definitive RRMS or PPMS disease course. Results were confirmed by quantitative reverse transcriptase polymerase chain reaction, immunocytochemistry and western blot of cultured cells, and validated in human brain specimens.

RESULTS

We identified common and unique oligodendrocyte genes for each treatment group. Exposure to CSF from PPMS uniquely induced branching of cultured progenitors and related transcriptional changes, including upregulation (P<0.05) of the adhesion molecule GALECTIN-3/Lgals3, which was also detected at the protein level in brain specimens from PPMS patients. This pattern of gene expression was distinct from the transcriptional programme of oligodendrocyte differentiation during development.

CONCLUSIONS

Despite evidence of morphological differentiation induced by exposure to CSF of PPMS patients, the overall transcriptional response elicited in cultured OPCs was consistent with the activation of an aberrant transcriptional programme.