Microarray analysis of autoimmune diseases by machine learning procedures

Systemic lupus in the era of machine learning medicine

Artificial intelligence and machine learning applications are emerging as transformative technologies in medicine. With greater access to a diverse range of big datasets, researchers are turning to these powerful techniques for data analysis. Machine learning can reveal patterns and interactions between variables in large and complex datasets more accurately and efficiently than traditional statistical methods. Machine learning approaches open new possibilities for studying SLE, a multifactorial, highly heterogeneous and complex disease. Here, we discuss how machine learning methods are rapidly being integrated into the field of SLE research. Recent reports have focused on building prediction models and/or identifying novel biomarkers using both supervised and unsupervised techniques for understanding disease pathogenesis, early diagnosis and prognosis of disease. In this review, we will provide an overview of machine learning techniques to discuss current gaps, challenges and opportunities for SLE studies. External validation of most prediction models is still needed before clinical adoption. Utilisation of deep learning models, access to alternative sources of health data and increased awareness of the ethics, governance and regulations surrounding the use of artificial intelligence in medicine will help propel this exciting field forward.

A systematic review of the applications of artificial intelligence and machine learning in autoimmune diseases

Autoimmune diseases are chronic, multifactorial conditions. Through machine learning (ML), a branch of the wider field of artificial intelligence, it is possible to extract patterns within patient data, and exploit these patterns to predict patient outcomes for improved clinical management. Here, we surveyed the use of ML methods to address clinical problems in autoimmune disease. A systematic review was conducted using MEDLINE, embase and computers and applied sciences complete databases. Relevant papers included "machine learning" or "artificial intelligence" and the autoimmune diseases search term(s) in their title, abstract or key words. Exclusion criteria: studies not written in English, no real human patient data included, publication prior to 2001, studies that were not peer reviewed, non-autoimmune disease comorbidity research and review papers. 169 (of 702) studies met the criteria for inclusion. Support vector machines and random forests were the most popular ML methods used. ML models using data on multiple sclerosis, rheumatoid arthritis and inflammatory bowel disease were most common. A small proportion of studies (7.7% or 13/169) combined different data types in the modelling process. Cross-validation, combined with a separate testing set for more robust model evaluation occurred in 8.3% of papers (14/169). The field may benefit from adopting a best practice of validation, cross-validation and independent testing of ML models. Many models achieved good predictive results in simple scenarios (e.g. classification of cases and controls). Progression to more complex predictive models may be achievable in future through integration of multiple data types.

Gender-specific associations between polymorphisms in the Toll-like receptor (TLR) genes and lung function among workers in swine operations

Workers in swine operations are exposed to dust, bacteria, and virus, and are at increased risk of respiratory problems. Toll-like receptors (TLR) play an important role in human immune responses to respiratory hazards. Worker gender and age may significantly modify the involvement of TLR in the etiology of these respiratory outcomes. The aim of this study was to investigate whether modification effects of worker gender and age altered associations between polymorphisms in the TLR genes and lung function. This study included 374 full-time workers from large swine operations from Saskatchewan. Information on demography, lifestyle, pulmonary function,  and blood samples were obtained. Multiple linear regression and decision tree model were used in the analysis. Among females aged <45.8 years, workers with polymorphisms of rs4696480 in the TLR2 gene exhibited markedly better lung function than workers with wild-type. These associations were not observed among female workers aged ≥45.8 years and males. Among males, workers with polymorphisms of rs187084 in the TLR9 gene displayed significantly lower lung function than those with wild-type. This male-specific association was not dependent on worker age. This is the first study to report gender-specific correlations between lung function and polymorphisms of TLR genes, and modification effects of worker age on these associations, suggesting the importance of considering gender and age in genetic association studies of airway diseases due to exposure of high concentration of respiratory hazards.

A critical role of E2F transcription factor 2 in proinflammatory cytokines-dependent proliferation and invasiveness of fibroblast-like synoviocytes in rheumatoid Arthritis

As a transcription factor, E2F2 participates in regulation of numerous genes. To investigate the role and mechnism of E2F2 in RA, expression of E2F2 in synovial tissue was detected. Proliferation, invasion, and secretion of inflammatory cytokines were measured after E2F2 was knocked-down in RASFs by siRNA transfection. Induction of TNF-α, IL-6, and LPS on expression and nuclear translocation of E2F2, and signal pathways involved in the process were tested. ChIP was used to investigate direct binding of NF-кB to the promoter of E2F2, and E2F2 to the promoter of IL-6. The correlation between mRNA levels of E2F2 and IL-6 or TNF-α in secreted in supernatant of RASFs were also investigated. As a result, silencing E2F2 could inhibit the proliferation and invasion of RASFs. LPS, IL-6 can stimulate the expression of E2F2 in RASFs both via the NF-кB pathway, while TNF-α via the ERK pathway. TNF-α can facilitate the nuclear translocation of E2F2 and TNF-α can bind to promoter of E2F2, and then E2F2 can bind to the promoter of IL-6 directly. Significant correlations was found between levels of E2F2 and IL-6/TNF-α in synoviocytes of RA patients. Our findings indicate that E2F2 may play an important role in pathogenesis of RA.

Endothelial progenitor dysfunction associates with a type I interferon signature in primary antiphospholipid syndrome

OBJECTIVES

Patients with antiphospholipid syndrome (APS) are at risk for subclinical endothelial injury, as well as accelerated atherosclerosis. In the related disease systemic lupus erythematosus, there is a well-established defect in circulating endothelial progenitors, which leads to an accrual of endothelial damage over time. This defect has been at least partially attributed to exaggerated expression of type I interferons (IFNs). We sought to determine whether these pathways are important in APS.

METHODS

We studied 68 patients with primary APS. Endothelial progenitors were assessed by flow cytometry and functional assay. Type I IFN activity was determined by a well-accepted bioassay, while peripheral blood mononuclear cells were scored for expression of IFN-responsive genes.

RESULTS

Endothelial progenitors from patients with APS demonstrated a marked defect in the ability to differentiate into endothelial cells, a phenotype which could be mimicked by treating control progenitors with APS sera. Elevated type I IFN activity was detected in the circulation of patients with APS (a finding that was then replicated in an independent cohort). While IgG depletion from APS sera did not rescue endothelial progenitor function, the dysfunction was successfully reversed by a type I IFN receptor-neutralising antibody.

CONCLUSIONS

We describe, for the first time to our knowledge, an IFN signature in primary APS and show that this promotes impaired endothelial progenitor function. This work opens the door to novel approaches that may mitigate vascular damage in APS, such as anti-IFN drugs.

Voxel-Based Diagnosis of Alzheimer's Disease Using Classifier Ensembles

Functional magnetic resonance imaging (fMRI) is one of the most promising noninvasive techniques for early Alzheimer's disease (AD) diagnosis. In this paper, we explore the application of different machine learning techniques to the classification of fMRI data for this purpose. The functional images were first preprocessed using the statistical parametric mapping toolbox to output individual maps of statistically activated voxels. A fast filter was applied afterwards to select voxels commonly activated across demented and nondemented groups. Four feature ranking selection techniques were embedded into a wrapper scheme using an inner-outer loop for the selection of relevant voxels. The wrapper approach was guided by the performance of six pattern recognition models, three of which were ensemble classifiers based on stochastic searches. Final classification performance was assessed from the nested internal and external cross-validation loops taking several voxel sets ordered by importance. Numerical performance was evaluated using statistical tests, and the best combination of voxel selection and classification reached a 97.14% average accuracy. Results repeatedly pointed out Brodmann regions with distinct activation patterns between demented and nondemented profiles, indicating that the machine learning analysis described is a powerful method to detect differences in several brain regions between both groups.

Type 1 diabetes in mice and men: gene expression profiling to investigate disease pathogenesis

Type 1 diabetes (T1D) is a complex polygenic disease that is triggered by various environmental factors in genetically susceptible individuals. The emphasis placed on genome-wide association studies to explain the genetics of T1D has failed to advance our understanding of T1D pathogenesis or identify biomarkers of disease progression or therapeutic targets. Using the nonobese diabetic (NOD) mouse model of T1D and the non-disease prone congenic NOD.B10 mice, our laboratory demonstrated striking tissue-specific and age-dependent changes in gene expression during disease progression. We established a "roadmap" of differential gene expression and used this to identify candidate genes in mice (and human orthologs) that play a role in disease pathology. Here, we describe two genes, Deformed epidermal autoregulatory factor 1 (Deaf1) and Adenosine A1 receptor (Adora1), that are differentially expressed and alternatively spliced in the pancreatic lymph nodes or islets of NOD mice and T1D patients to form dominant-negative non-functional isoforms. Loss of Deaf1 function leads to reduced peripheral tissue antigen expression in lymph node stromal cells and may contribute to a breakdown in peripheral tolerance, while reduced Adora1 function results in an early intrinsic alpha cell defect that may explain the hyperglucagonemia and resulting beta cell stress observed prior to the onset of diabetes. Remarkably, both genes were also alternatively spliced in the same tissues of auto-antibody positive prediabetic patients, and these splicing events resulted in similar downstream effects as those seen in NOD mice. These findings demonstrate the value of gene expression profiling in studying disease pathogenesis in T1D.

Gene expression in B-1 cells from lupus-prone mice

New Zealand Black X New Zealand White F1 [(NZB/NZW)F1] mice develop an autoimmune condition with similarities to human systemic lupus erythematosus (SLE). In this study, we demonstrate that B-1 cells, which have previously been reported to be involved in several autoimmune diseases, have altered gene expression in these mice. RNA was extracted from purified B-1 cells of disease-free C57BL/6 mice and lupus-prone (NZB/NZW)F1 mice. Gene expression was analysed using DNA microarray techniques and validated by real time reverse transcriptase polymerase chain reaction (RT-PCR). In (NZB/NZW)F1 mice, some genes had altered expression patterns compared to disease-free controls. Specifically, the upregulation of Ifitm1, Pvrl2 and Ifi202b and downregulation of Trp53bp1 mRNA were observed in (NZB/NZW)F1 mice. These genes are known to be associated with autoimmune diseases. This pattern of gene expression in B-1 cells could understanding of the pathogenesis of SLE. Thus, it is reasonable to hypothesise that the altered gene expression observed in B-1 cells in our experimental model is important for SLE prognosis and therapy, and these implications are discussed herein.

MicroRNA-let-7a promotes E2F-mediated cell proliferation and NFκB activation in vitro

Epigenetic factors, including altered microRNA (miRNA) expression, may contribute to aberrant immune cell function in systemic lupus erythematosus (SLE). MiRNA-let-7a (let-7a) has been shown to directly alter cell cycle progression and proinflammatory cytokine production. Due to the crucial role of let-7a in cell division and inflammation, we investigated let-7a-mediated proliferation and NFκB translocation in J774A.1 macrophages and MES 13 mesangial cells in vitro. In immune-stimulated cells transfected with let-7a, cell proliferation was significantly increased over time. There was a significant increase in the number of immune-stimulated cells in S and G2 phases. Immune-stimulated cells overexpressing let-7a had increased nuclear translocation of NFκB. Bioinformatical analysis revealed that the E2F family, critical regulators of the G1-S transition, has potential binding sites for let-7a in their mRNA transcripts. Let-7a overexpression significantly increased the expression of the cell cycle activator E2F2 and increased retinoblastoma protein (Rb) phosphorylation in immune-stimulated cells. The cell cycle inhibitor E2F5 was significantly decreased in let-7a-transfected cells that were immune-stimulated. Bioinformatical analysis revealed E2F2 and NFκB are transcription factors predicted to regulate the let-7a promoter. We analyzed transcriptional regulation of let-7a by real-time RT-PCR using chromatin immunoprecipitation with E2F2 and NFκB antibodies. There was an increase in E2F2 and NFκB binding in DNA enriched for the let-7a promoter in immune-stimulated cells. Silencing E2F2 or NFκB significantly decreased let-7a expression and IL-6 production in immune-stimulated cells. Taken together, our results suggest that overexpression of let-7a may contribute to hyperplasia and the proinflammatory response in SLE.

Ensemble transcript interaction networks: a case study on Alzheimer's disease

Systems biology techniques are a topic of recent interest within the neurological field. Computational intelligence (CI) addresses this holistic perspective by means of consensus or ensemble techniques ultimately capable of uncovering new and relevant findings. In this paper, we propose the application of a CI approach based on ensemble Bayesian network classifiers and multivariate feature subset selection to induce probabilistic dependences that could match or unveil biological relationships. The research focuses on the analysis of high-throughput Alzheimer's disease (AD) transcript profiling. The analysis is conducted from two perspectives. First, we compare the expression profiles of hippocampus subregion entorhinal cortex (EC) samples of AD patients and controls. Second, we use the ensemble approach to study four types of samples: EC and dentate gyrus (DG) samples from both patients and controls. Results disclose transcript interaction networks with remarkable structures and genes not directly related to AD by previous studies. The ensemble is able to identify a variety of transcripts that play key roles in other neurological pathologies. Classical statistical assessment by means of non-parametric tests confirms the relevance of the majority of the transcripts. The ensemble approach pinpoints key metabolic mechanisms that could lead to new findings in the pathogenesis and development of AD.

Identification of a biomarker panel for colorectal cancer diagnosis

BACKGROUND

Malignancies arising in the large bowel cause the second largest number of deaths from cancer in the Western World. Despite progresses made during the last decades, colorectal cancer remains one of the most frequent and deadly neoplasias in the western countries.

METHODS

A genomic study of human colorectal cancer has been carried out on a total of 31 tumoral samples, corresponding to different stages of the disease, and 33 non-tumoral samples. The study was carried out by hybridisation of the tumour samples against a reference pool of non-tumoral samples using Agilent Human 1A 60-mer oligo microarrays. The results obtained were validated by qRT-PCR. In the subsequent bioinformatics analysis, gene networks by means of Bayesian classifiers, variable selection and bootstrap resampling were built. The consensus among all the induced models produced a hierarchy of dependences and, thus, of variables.

RESULTS

After an exhaustive process of pre-processing to ensure data quality--lost values imputation, probes quality, data smoothing and intraclass variability filtering--the final dataset comprised a total of 8, 104 probes. Next, a supervised classification approach and data analysis was carried out to obtain the most relevant genes. Two of them are directly involved in cancer progression and in particular in colorectal cancer. Finally, a supervised classifier was induced to classify new unseen samples.

CONCLUSIONS

We have developed a tentative model for the diagnosis of colorectal cancer based on a biomarker panel. Our results indicate that the gene profile described herein can discriminate between non-cancerous and cancerous samples with 94.45% accuracy using different supervised classifiers (AUC values in the range of 0.997 and 0.955).

Peakbin selection in mass spectrometry data using a consensus approach with estimation of distribution algorithms

Progress is continuously being made in the quest for stable biomarkers linked to complex diseases. Mass spectrometers are one of the devices for tackling this problem. The data profiles they produce are noisy and unstable. In these profiles, biomarkers are detected as signal regions (peaks), where control and disease samples behave differently. Mass spectrometry (MS) data generally contain a limited number of samples described by a high number of features. In this work, we present a novel class of evolutionary algorithms, estimation of distribution algorithms (EDA), as an efficient peak selector in this MS domain. There is a trade-of f between the reliability of the detected biomarkers and the low number of samples for analysis. For this reason, we introduce a consensus approach, built upon the classical EDA scheme, that improves stability and robustness of the final set of relevant peaks. An entire data workflow is designed to yield unbiased results. Four publicly available MS data sets (two MALDI-TOF and another two SELDI-TOF) are analyzed. The results are compared to the original works, and a new plot (peak frequential plot) for graphically inspecting the relevant peaks is introduced. A complete online supplementary page, which can be found at http://www.sc.ehu.es/ccwbayes/members/ruben/ms, includes extended info and results, in addition to Matlab scripts and references.