Improved Regularized Multi-class Logistic Regression for Gene Classification with Optimal Kernel PCA and HC Algorithm.

The gene expression profile of tissue in the colonic mucosa is highly altered in ulcerative colitis (UC) compared with healthy tissue. Analyses of microarray data taken from colonic mucosa biopsies shows upregulation of immune-related genes and may help to indicate molecular pathways that may be targeted for therapeutic intervention. At the same time, genome-wide association studies (GWAS) of ulcerative colitis have identified genetic susceptibility loci highly enriched in immune-related genes as well. We sought to establish the extent of overlap between pathways upregulated in gene expression with genetic susceptibility loci in order to generate a list of core gene sets that can guide functional investigation of UC pathobiology. We conducted a Gene Set Enrichment Analysis (GSEA) in 3 sets of colonic biopsy studies assayed on the Affymetrix U133 Plus 2 array that were publically available through the Gene Expression Omnibus. We also performed a GSEA on genotype data from 8 European-ancestry cohorts consisting of 5584 cases and 11587 controls using a hypergeometric test with the same gene set database of manually-curated pathways from BioCarta, Reactome, and Kyoto Encyclopedia of Genes and Genomes. GSEA revealed that all 3 gene expression datasets share 87 gene sets upregulated in UC lesions and 8 gene sets downregulated (false discovery rate <0.05). The upregulated pathways were dominated by gene sets involved in immune function and signaling, as well as the control of mitosis. GSEA in the GWAS study produced 56 gene sets, 33 of which were concordant with the gene-expression-derived pathways (hypergeometric P = 1.54 × 10−18 for this overlap). The identified pathways were dominated by immune system function, control of mitosis, signaling pathways, oxidative phosphorylation, ubiquitin/proteasome, and extracellular matrix interaction. This study reveals that a concordant set of pathways are altered both at the transcriptome level as well as having risk-conferring genetic variation. These common pathways can serve as a useful guide to prioritizing targets for therapeutic development or establishing personalized treatment of ulcerative colitis.

In this study, gene expression data of osteosarcoma (OSA) were analyzed to identify metastasis-related biological pathways. Four gene expression data sets (GSE21257, GSE9508, GSE49003 and GSE66673) were downloaded from Gene Expression Omnibus (GEO). An analysis of differentially expressed genes (DEGs) was performed using the Significance Analysis of Microarray (SAM) method. Gene expression levels were converted into scores of pathways by the Functional Analysis of Individual Microarray Expression (FAIME) algorithm and the differentially expressed pathways (DEPs) were then disclosed by a t-test. The distinguishing and prediction ability of the DEPs for metastatic and non-metastatic OSA was further confirmed using the principal component analysis (PCA) method and 3 gene expression data sets (GSE9508, GSE49003 and GSE66673) based on the support vector machines (SVM) model. A total of 616 downregulated and 681 upregulated genes were identified in the data set, GSE21257. The DEGs could not be used to distinguish metastatic OSA from non-metastatic OSA, as shown by PCA. Thus, an analysis of DEPs was further performed, resulting in 14 DEPs, such as NRAS signaling, Toll-like receptor (TLR) signaling, matrix metalloproteinase (MMP) regulation of cytokines and tumor necrosis factor receptor-associated factor (TRAF)-mediated interferon regulatory factor 7 (IRF7) activation. Cluster analysis indicated that these pathways could be used to distinguish between metastatic OSA from non-metastatic OSA. The prediction accuracy was 91, 66.7 and 87.5% for the data sets, GSE9508, GSE49003 and GSE66673, respectively. The results of PCA further validated that the DEPs could be used to distinguish metastatic OSA from non-metastatic OSA. On the whole, several DEPs were identified in metastatic OSA compared with non-metastatic OSA. Further studies on these pathways and relevant genes may help to enhance our understanding of the molecular mechanisms underlying metastasis and may thus aid in the development of novel therapies.

Periodontitis and oral squamous cell carcinoma (OSCC) are prevalent oral diseases with distinct etiologies, yet they share certain molecular and biological characteristics. Gene expression datasets from the gene expression omnibus (GEO) repository (GSE30784 for OSCC and GSE10334 for periodontitis) were analyzed. Data preprocessing and differential gene expression analysis using GEO2R identified common differentially expressed genes (DEGs), and FunRich software facilitated the construction of a protein-protein interaction (PPI) network on the STRING database. Cytoscape, coupled with the CytoHubba plugin, identified Cluster of Differentiation 19 (CD19) and Von Willebrand Factor (VWF) as the top hub genes, with Complement C3 (C3) also highly ranked. Functional enrichment analysis highlighted pathways such as the B-cell receptor signaling pathway, complement and coagulation cascades, and hematopoietic cell lineage. Additionally, miRNet analysis identified key miRNAs, including hsa-mir-26a-5p, hsa-mir-129-2-3p, and hsa-mir-27a-3p, associated with these pathways. These findings suggested an association of molecular mechanisms between periodontitis and OSCC, with identified hub genes and miRNAs potentially serving as therapeutic targets.

Background: Non-obstructive azoospermia (NOA) is the most severe form of male infertility. Currently, the molecular mechanisms underlying NOA pathology have not yet been elucidated. Hence, elucidation of the mechanisms of NOA and exploration of potential biomarkers are essential for accurate diagnosis and treatment of this disease. In the present study, we aimed to screen for biomarkers and pathways involved in NOA and reveal their potential molecular mechanisms using integrated bioinformatics.Methods: We downloaded two gene expression datasets from the Gene Expression Omnibus (GEO) database. Differentially expressed genes (DEGs) in NOA and matched the control group tissues were identified using the limma package in R software. Subsequently, Gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), gene set enrichment analysis (GSEA), protein-protein interaction (PPI) network, gene-microRNAs network, and transcription factor (TF)-hub genes regulatory network analyses were performed to identify hub genes and associated pathways. Finally, we conducted immune infiltration analysis using CIBERSORT to evaluate the relationship between the hub genes and the NOA immune infiltration levels.Results: We identified 698 common DEGs, including 87 commonly upregulated and 611 commonly downregulated genes in the two datasets. GO analysis indicated that the most significantly enriched gene was protein polyglycylation, and KEGG pathway analysis revealed that the DEGs were most significantly enriched in taste transduction and pancreatic secretion signaling pathways. GSEA showed that DEGs affected the biological functions of the ribosome, focaladhesion, and protein_expor. We further identified the top 31 hub genes from the PPI network, and friends analysis of hub genes in the PPI network showed that NR4A2 had the highest score. In addition, immune infiltration analysis found that CD8+ T cells and plasma cells were significantly correlated with ODF3 expression, whereas naive B cells, plasma cells, monocytes, M2 macrophages, and resting mast cells showed significant variation in the NR4A2 gene expression group, and there were differences in T cell regulatory immune cell infiltration in the FOS gene expression groups.Conclusion: The present study successfully constructed a regulatory network of DEGs between NOA and normal controls and screened three hub genes using integrative bioinformatics analysis. In addition, our results suggest that functional changes in several immune cells in the immune microenvironment may play an important role in spermatogenesis. Our results provide a novel understanding of the molecular mechanisms of NOA and offer potential biomarkers for its diagnosis and treatment.

Angiitis, also known as vasculitis, is a chronic inflammatory disease characterized by the infiltration of inflammatory cells in surroundings of blood vessels, accompanied by vascular damage including fibrin deposition, collagen fiber degeneration, myocyte, and endotheliocyte necrosis. This work aimed to perform an integrated bioinformatic analysis of three data sets concerning vasculitis to explore and examine the potential diagnostic and therapeutic makers contributing to illuminating the pathomechanisms of vasculitis. We collected three sets of gene expression data designed by dual-channel method from Gene Expression Omnibus, which were based on the same platform (Agilent-014850 Whole Human Genome Microarray 4x44K G4112F). The meta-analysis was used to analyze the gene expression profiles and screen the differentially expressed genes followed by functional features identification. Subsequently, a protein-protein interaction and transcriptional regulation network were conducted for further investigation of expression mechanisms of vasculitis. Totally, 73 consistently upregulated genes, 49 consistently downregulated genes, and 26 genes with different expression directions were identified. Functional enrichment and transcription regulation analysis suggested upregulated genes (PPBP, PLAU, and HIST1H2BH) and downregulated genes such as IL23A gene were predominately associated with immune responses and cytokine receptors function. In addition, specific cancer-related genes such as MRVI1 was also extracted and considered as promising biomarkers of the development and progression of vasculitis. This study established an integrated meta-analysis approach and identified novel biomarkers involved in vasculitis, which further facilitate to explore and unravel the etiopathogenesis of vasculitis.

Gene expression variation is a quantitative trait that drives phenotypic diversity across populations. On a cellular level, gene expression is an intermediate phenotype between stored genetic information and the functional utilization of this information within the cell. Through Genome Wide Association Studies(GWAS), thousands of genetic polymorphisms associated with numerous diseases have been identified. These have provided many novel insights into the disrupted biological processes that drive the etiology of various health conditions. Expression Quantitative Trait Loci (eQTL) provide an additional layer of biological information about the physiological impact of common genetic variants. Therefore, the study of the genetic regulation of gene expression (eQTL studies) has been useful both in the validation and functional characterisation of GWAS polymorphisms. This has contributed to a better understanding of the precise molecular processes that contribute to the development of disease. Global transcriptomic analyses have provided as greater insight into the level of complexity that drives biological systems. Transcriptomic data are often comprised of gene regulatory and co-expression networks, an emergent property of transcriptomic and other nomic data. These networks within each nomics fields interact with each other to further add layers of complexity that drive biological systems. Variation contained with gene expression datasets can, therefore, provide detail into the flow of information through these biological systems and how these can be influenced by genetic polymorphisms. Transcriptomic variation is highly influenced by genetic and environmental factors. Genetic regulation of gene expression represents, with some exceptions, fixed regulatory points that strictly control the expression of genes. Variance attributed to environmental effects, on the other hand, are often biological responses to specific stimuli. The dissection of the genetic and environmental influences on expression levels will help to form a baseline upon which network models can be built to disseminate the biological flow of information in healthy, latent or disease groups. This thesis will detail both methodological methods to clean data, and statistical approaches analyze the complexities found within the variance of transcriptomic data. The focus of this thesis is the dissection of three major influences of gene expression variability: technical artifacts, environmental and genetic variation. Using statistical and quantitative genetic techniques on array-based genotype and gene expression datasets, this thesis examines: 1. The use of Principal Components Analysis (PCA) to identify and correct for known batch effects 2. Season variation as a pervasive environmental contributor to gene expression variation 3. The genetic contribution driving robust gene co-expression modules The Brisbane Systems Genetics Study (BSGS) is comprised of both unrelated and related individuals and was used throughout these three studies. The Center for Healthy Discovery and Well Being {CHDWB} cohort was used as a replication study and the Multiple Tissue Human Expression Resource (MuTHER) cohort was employed to examine tissue-specific effects. The first chapter provides a technical methodological analysis of the batch effect correction technique PCA. Batch effects have a large impact on gene expression variability, often creating artificial systematic trends. By decomposing the data in Principal Components (PCs) we were able to quantify the degree and distribution of technical artifacts within gene expression datasets and determine the effectiveness of this correction method. The second chapter examines the influence of pervasive macro-environmental factors on gene expression datasets and provides a statistical framework to identify seasonal variation. Since datasets are often collected over time, samples may contain seasonal trends in gene expression that are environmentally driven and are not regarded as technical artifacts. By using loess decomposition and cosinor regression, 74 transcripts with a significant season trend were identified independently of seasonal variation in blood cell count. Chapter three examines the genetic contribution to gene expression covariance between transcripts, called Blood Informative Transcripts (BITs) comprising of nine modules that have been previously identified and validated. Using quantitative genetic techniques, the genetic and environmental components driving phenotypic correlations for BIT transcripts were quantified. When compared to 10,000 bootstrap permutations of random probes the BITS demonstrate significant genetic correlation (average 0.63 across all BITs) and an average genetic contribution to phenotypic correlations of 0.42. The high degree of genetic correlation demonstrates a strong genetic framework regulating the expression of BITs modules. This chapter also examines the presence of this replicated modules in three separate tissue types, identifying several tissue-specific coexpression modules. Overall, this thesis explores the landscape of gene expression variability. A methodological framework for the identification of technical artifacts and seasonal effects is investigated and the genetic architecture driving transcriptomic co-expression is characterized.

BackgroundThe survival of patients with breast cancer is highly sporadic, from a few months to more than 15 years. In recent studies, the gene expression profiling of tumors has been used as a promising means of predicting prognosis factors.MethodsIn this study, we used gene expression datasets of tumors to identify prognostic factors in breast cancer. We conducted log-rank tests and used unsupervised clustering methods to find reciprocally expressed gene sets associated with worse survival rates. Prognosis prediction scores were determined as the ratio of gene expressions.ResultsAs a result, four prognosis prediction gene set modules were constructed. The four prognostic gene sets predicted worse survival rates in three independent gene expression data sets. In addition, we found that cancer patient with poor prognosis, i.e., triple-negative cancer, HER2-enriched, TP53 mutated and high-graded patients had higher prognosis prediction scores than those with other types of breast cancer.ConclusionsIn conclusion, based on a gene expression analysis, we suggest that our well-defined scoring method of the prediction of survival outcome may be useful for developing prognostic factors in breast cancer.Electronic supplementary materialThe online version of this article (doi:10.1186/s12911-016-0292-5) contains supplementary material, which is available to authorized users.

Constantly improving gene expression profiling technologies are expected to provide understanding and insight into cancer-related cellular processes. Gene expression data is also expected to significantly aid in the development of efficient cancer diagnosis and classification platforms. In this work we examine three sets of gene expression data measured across sets of tumor(s) and normal clinical samples: The first set consists of 2,000 genes, measured in 62 epithelial colon samples (Alon et al., 1999). The second consists of approximately equal to 100,000 clones, measured in 32 ovarian samples (unpublished extension of data set described in Schummer et al. (1999)). The third set consists of approximately equal to 7,100 genes, measured in 72 bone marrow and peripheral blood samples (Golub et al, 1999). We examine the use of scoring methods, measuring separation of tissue type (e.g., tumors from normals) using individual gene expression levels. These are then coupled with high-dimensional classification methods to assess the classification power of complete expression profiles. We present results of performing leave-one-out cross validation (LOOCV) experiments on the three data sets, employing nearest neighbor classifier, SVM (Cortes and Vapnik, 1995), AdaBoost (Freund and Schapire, 1997) and a novel clustering-based classification technique. As tumor samples can differ from normal samples in their cell-type composition, we also perform LOOCV experiments using appropriately modified sets of genes, attempting to eliminate the resulting bias. We demonstrate success rate of at least 90% in tumor versus normal classification, using sets of selected genes, with, as well as without, cellular-contamination-related members. These results are insensitive to the exact selection mechanism, over a certain range.

BackgroundThis study aimed to identify potential novel therapeutic targets for nasopharyngeal carcinoma (NPC) by identifying aberrantly methylated-differentially expressed genes (DEGs) and pathways based on a comprehensive bioinformatics analysis.MethodsEight gene expression data sets and 2 methylation microarray data sets that included NPC and control groups from the Gene Expression Omnibus were identified. Meta-analyses of the DEGs were performed using the online analysis database “NetworkAnalyst”. Aberrantly methylated gene loci were obtained from the GEO2R. Aberrantly methylated DEGs were obtained from Venn diagrams. The enrichment analysis was carried out on the “Metascape” website, and the protein-protein interaction (PPI) network construction, network analysis, and visualization of the analysis results were carried out on the “String” website using “Cytoscape” software.ResultsIn total, 544 hypomethylation high-expression genes and 164 hypermethylation low-expression genes were obtained. The enrichment and PPI network analyses suggested that several pathways and hub genes with abnormal gene expression accompanied by methylation change, including inositol-trisphosphate 3-kinase B (ITPKB), G protein subunit beta 5 (GNB5), FYN proto-oncogene, Src family tyrosine kinase (FYN), LCK proto-oncogene, Src family tyrosine kinase (LCK), nuclear factor of activated T cells 1 (NFATC1), GNAS complex locus (GNAS), protein kinase C beta (PRKCB), zeta chain of T cell receptor associated protein kinase 70 (ZAP70), lysophosphatidic acid receptor 1 (LPAR1), protein kinase C epsilon (PRKCE), tumor protein p53 (TP53), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), fibronectin 1 (FN1), cyclin D1 (CCND1), vascular endothelial growth factor A (VEGFA), HRas proto-oncogene, GTPase (HRAS), signal transducer and activator of transcription 3 (STAT3), fibroblast growth factor 2 (FGF2), amyloid beta precursor protein (APP), and matrix metallopeptidase 2 (MMP2), may be related to the occurrence of nasopharyngeal carcinoma .ConclusionsThe identification of novel and important pathways and hub genes and their roles in the occurrence and development of NPC will guide clinical research and the development of pharmaceutical targets.

Uveal melanoma (UM) is the most common intraocular tumor worldwide. We proposed to identifya vital gene signature that has prognostic value for UM metastasis. For this purpose, we obtained a published DNA methylation and gene expression data set associated with UM from the Gene Expression Omnibus. The genes whose aberrant expression significantly associated with UM patients' metastasis-free survival (MFS) were identified by applying a univariate Cox proportional hazards model to the gene expression data set followed by a robust likelihood-based survival analysis to screen the optimal prognostic gene signatures (PGS). A formula for calculating the risk score that represents UM metastasis risk was constructed by including the PGSs' expression values weighted by their regression coefficients, which were obtained by a multivariate Cox regression analysis. As a result, aberrant expression of 2884 genes were found to be significantly associated with UM patients' MFS, which were referred to as MFSGs, and 11 out of those MFSGs, GJC1, TCEA1, MFSD3, FAF2, TLCD1, GPAA1, CYC1, ASAP1, JPH1, LDB3, and KDELR3, were identified as PGSs through which we could accurately separate UM samples with shorter MFS from those with longer MFS. By combining the DNA methylation data set and MFSGs, we further identified 265 MFSGs, which contained CpG sites that significantly hyper- or hypo-methylated in UM samples compared with control samples. Functional enrichment analysis and pathway crosstalk analysis of those genes indicated significant enrichment of cancer-related pathways. In conclusion, we identified an 11-gene-based prognostic signature and several gene biomarkers for UM metastasis, which should be helpful for selecting an appropriate treatment method for specific patients with UM.

Gene expression data are widely used in diagnosing diseases and identifying promising genes with the advancement in computational tools in biology. Gene Expression Omnibus (GEO) datasets provide the gene expression data for various diseases and disorders. For Bipolar Disorder, GSE46449 was obtained from the NCBI data repository. This study aimed to classify control (Normal) and case (Disordered) individuals from samples using Machine Learning (ML)/Deep Learning (DL) models. The preprocessing involved the removal of null values and normalization of gene expression values using R. The second step focussed on the selection of optimal features/genes from the gene expression dataset. The Pearson Correlation Coefficient (PCC) along with Principal Component Analysis (PCA) were used for feature selection. The samples were then classified using ML/DL models. A Multi-Layer Perceptron (MLP) was used to validate the optimal feature set to classify healthy and disordered individuals. The proposed Bipolar Disorder Preprocessing Framework (BiPoP) was validated for its targeted use, highlighting its multifunctional and fine-tuned approach to preprocessing and achieving a classification accuracy of 98.9%.

Gene expression data using retrieved ovarian cancer (OC) samples were used to identify genes of interest and a support vector machine (SVM) classifier was subsequently established to predict the recurrence of OC. Three datasets (GSE17260, GSE44104 and GSE51088) investigating OC gene expression were downloaded from the Gene Expression Omnibus. Differentially expressed genes (DEGs) in samples from patients with non-recurrent and recurrent OC were revealed via a homogeneity test and quality control analysis. A protein-protein interaction (PPI) network was subsequently established for the DEGs using data from Biological General Repository for Interaction Datasets, Human Protein Reference Database and Database of Interacting Proteins. Degrees of interaction and betweenness centrality (BC) scores were calculated for each node in the PPI network. The top 100 genes ranked by BC scores were selected to identify feature genes via recursive feature elimination using the GSE17260 dataset. Following this, a SVM classifier was constructed and further validated using the GSE44104 and GSE51088 datasets and independent gene expression data obtained from the Cancer Genome Atlas (TCGA). A total of 639 DEGs were identified from the three gene expression datasets, and a PPI network including 249 nodes and 354 edges was constructed. A SVM classifier consisting of 39 feature genes (including cullin 3, mouse double minute 2 homolog, aurora kinase A, WW domain containing oxidoreducatase, large tumor suppressor kinase 2, sirtuin 6, staphylococcal nuclease and tudor domain containing 1, leucine rich repeats and immunoglobulin like domains 1 and aurora kinase 1 interacting protein 1) was subsequently constructed. The prediction accuracies of the SVM classifier for GSE17260, GSE44104 and GSE51088 datasets as well as data downloaded from TCGA were revealed to be 92.7, 93.3, 96.6 and 90.4%, respectively. Furthermore, the results of the present study revealed that patients with predicted non-recurrent OC survived significantly longer compared with the patients with predicted recurrent OC (P=6.598×10−6). A SVM classifier consisting of 39 feature genes was established for predicting the recurrence and prognosis of OC. Therefore, the results of the present study suggested that the 39 feature genes may serve important roles in the development of OC and may represent therapeutic biomarkers of OC.

With the continuous expansion of available transcriptomic data like gene expression, deep learning techniques are becoming more and more valuable in analyzing and interpreting them. The National Center for Biotechnology Information Gene Expression Omnibus (GEO) encompasses approximately 5 million gene expression datasets from animal and human subjects. Unfortunately, the majority of them do not have a recorded timestamps, hindering the exploration of the behavior and patterns of circadian genes. Therefore, predicting the phases of these unordered gene expression measurements can help understand the behavior of the circadian genes, thus providing valuable insights into the physiology, behaviors, and diseases of humans and animals. In this paper, we propose a novel approach to predict the phases of the un-timed samples based on a deep neural network architecture. It incorporates the potential periodic oscillation information of the cyclic genes into the objective function to regulate the phase estimation. To validate our method, we use mouse heart, mouse liver and temporal cortex of human brain dataset. Through our experiments, we demonstrate the effectiveness of our proposed method in predicting phases and uncovering rhythmic pattern in circadian genes.

Protein kinase C (PKC) isozymes are key signal transducers involved in normal physiology and disease and have been widely implicated in cancer progression. Despite our extensive knowledge of the signaling pathways regulated by PKC isozymes and their effectors, there is essentially no information on how individual members of the PKC family regulate gene transcription. Here, we report the first PKC isozyme-specific analysis of global gene expression by microarray using RNAi depletion of diacylglycerol/phorbol ester-regulated PKCs. A thorough analysis of this microarray data revealed unique patterns of gene expression controlled by PKCα, PKCδ, and PKCε, which are remarkably different in cells growing in serum or in response to phorbol ester stimulation. PKCδ is the most relevant isoform in controlling the induction of genes by phorbol ester stimulation, whereas PKCε predominantly regulates gene expression in serum. We also established that two PKCδ-regulated genes, FOSL1 and BCL2A1, mediate the apoptotic effect of phorbol esters or the chemotherapeutic agent etoposide in prostate cancer cells. Our studies offer a unique opportunity for establishing novel transcriptional effectors for PKC isozymes and may have significant functional and therapeutic implications.

Our aim was to identify differentially expressed (DE) genes and biological processes that may help predict patient response to biologic agents for rheumatoid arthritis (RA). Using the INMEX (integrative meta-analysis of expression data) software tool, we performed a meta-analysis of publicly available microarray Gene Expression Omnibus (GEO) datasets that examined patient response to biologic therapy for RA. Three GEO datasets, containing 79 responders and 34 non-responders, were included in the meta-analysis. We identified 1,374 genes that were consistently differentially expressed in responders vs. non-responders (651 up-regulated and 723 down-regulated). The up-regulated gene with the smallest p value (p=0.000192) was ASCC2 (Activating Signal Cointegrator 1 Complex Subunit 2), and the up-regulated gene with the largest fold change (average log fold change=-0.75869, p=0.000206) was KLRC3 (Killer Cell Lectin-Like Receptor Subfamily C, Member 3). The down-regulated gene with the smallest p value (p=0.000195) was MPL (Myeloproliferative Leukemia Virus Oncogene). Among the 236 GO terms associated with the set of DE genes, the most significantly enriched was "CTP biosynthetic process" (GO:0006241; p=0.000454). Our meta-analysis identified genes that were consistently DE in responders vs. non-responders, as well as biological pathways associated with this set of genes. These results provide insight into the molecular mechanisms underlying responsiveness to biologic therapy for RA.