Traditional Differential Expression Analysis Research Articles

SOPHIE: Generative Neural Networks Separate Common and Specific Transcriptional Responses

Genome-wide transcriptome profiling identifies genes that are prone to differential expression (DE) across contexts, as well as genes with changes specific to the experimental manipulation. Distinguishing genes that are specifically changed in a context of interest from common differentially expressed genes (DEGs) allows more efficient prediction of which genes are specific to a given biological process under scrutiny. Currently, common DEGs or pathways can only be identified through the laborious manual curation of experiments, an inordinately time-consuming endeavor. Here we pioneer an approach, Specific cOntext Pattern Highlighting In Expression data (SOPHIE), for distinguishing between common and specific transcriptional patterns using a generative neural network to create a background set of experiments from which a null distribution of gene and pathway changes can be generated. We apply SOPHIE to diverse datasets including those from human, human cancer, and bacterial pathogen Pseudomonas aeruginosa. SOPHIE identifies common DEGs in concordance with previously described, manually and systematically determined common DEGs. Further molecular validation indicates that SOPHIE detects highly specific but low-magnitude biologically relevant transcriptional changes. SOPHIE’s measure of specificity can complement log2 fold change values generated from traditional DE analyses. For example, by filtering the set of DEGs, one can identify genes that are specifically relevant to the experimental condition of interest. Consequently, these results can inform future research directions. All scripts used in these analyses are available at https://github.com/greenelab/generic-expression-patterns. Users can access https://github.com/greenelab/sophie to run SOPHIE on their own data.

Using Machine Learning to Identify Biomarkers Affecting Fat Deposition in Pigs by Integrating Multisource Transcriptome Information

Fat deposition in pigs is not only closely related to pig production efficiency and pork quality but also an ideal model for human obesity. Transcriptome sequencing is widely used to study fat deposition. However, due to small sample sizes, high false positive rates, and poor consistency of results from different studies, new strategies are urgently needed. Machine learning, a new analysis method, can effectively fit complex data and accurately identify samples and genes. In this study, 36 samples of adipose tissue, muscle tissue, and liver tissue were collected from Songliao black pigs and Landrace pigs, and the mRNA of all the samples was sequenced. In addition, we collected transcriptome data for 64 samples in the GEO database from four different sources. After standardization and imputation of missing values in the data set comprising 100 samples, traditional differential expression analysis was carried out, and different numbers of expressed genes were selected as features for the training model of eight machine learning methods. In the 1000 replications of fourfold cross validation with 100 samples, AdaBoost performed best, with an average prediction accuracy greater than 93% and the highest mean area under the curve in predicting the high- and low-fat content groups among the eight ML methods. According to their performance-based ranks inferred by AdaBoost, 12 genes related to fat deposition were identified; among them, FASN and APOD were specifically expressed in adipose tissue, and APOA1 was specifically expressed in the liver, which could be important candidate biomarkers affecting fat deposition.

Gene co-expression changes underlying the functional connectomic alterations in Alzheimer\u2019s disease

BackgroundThere is growing evidence indicating that a number of functional connectivity networks are disrupted at each stage of the full clinical Alzheimer’s disease spectrum. Such differences are also detectable in cognitive normal (CN) carrying mutations of AD risk genes, suggesting a substantial relationship between genetics and AD-altered functional brain networks. However, direct genetic effect on functional connectivity networks has not been measured.MethodsLeveraging existing AD functional connectivity studies collected in NeuroSynth, we performed a meta-analysis to identify two sets of brain regions: ones with altered functional connectivity in resting state network and ones without. Then with the brain-wide gene expression data in the Allen Human Brain Atlas, we applied a new biclustering method to identify a set of genes with differential co-expression patterns between these two set of brain regions.ResultsDifferential co-expression analysis using biclustering method led to a subset of 38 genes which showed distinctive co-expression patterns between AD-related and non AD-related brain regions in default mode network. More specifically, we observed 4 sub-clusters with noticeable co-expression difference, where the difference in correlations is above 0.5 on average.ConclusionsThis work applies a new biclustering method to search for a subset of genes with altered co-expression patterns in AD-related default mode network regions. Compared with traditional differential expression analysis, differential co-expression analysis yielded many more significant hits with extra insights into the wiring mechanism between genes. Particularly, the differential co-expression pattern was observed between two sets of genes, suggesting potential upstream genetic regulators in AD development.

DEGnext: classification of differentially expressed genes from RNA-seq data using a convolutional neural network with transfer learning

BackgroundA limitation of traditional differential expression analysis on small datasets involves the possibility of false positives and false negatives due to sample variation. Considering the recent advances in deep learning (DL) based models, we wanted to expand the state-of-the-art in disease biomarker prediction from RNA-seq data using DL. However, application of DL to RNA-seq data is challenging due to absence of appropriate labels and smaller sample size as compared to number of genes. Deep learning coupled with transfer learning can improve prediction performance on novel data by incorporating patterns learned from other related data. With the emergence of new disease datasets, biomarker prediction would be facilitated by having a generalized model that can transfer the knowledge of trained feature maps to the new dataset. To the best of our knowledge, there is no Convolutional Neural Network (CNN)-based model coupled with transfer learning to predict the significant upregulating (UR) and downregulating (DR) genes from both trained and untrained datasets.ResultsWe implemented a CNN model, DEGnext, to predict UR and DR genes from gene expression data obtained from The Cancer Genome Atlas database. DEGnext uses biologically validated data along with logarithmic fold change values to classify differentially expressed genes (DEGs) as UR and DR genes. We applied transfer learning to our model to leverage the knowledge of trained feature maps to untrained cancer datasets. DEGnext’s results were competitive (ROC scores between 88 and 99%) with those of five traditional machine learning methods: Decision Tree, K-Nearest Neighbors, Random Forest, Support Vector Machine, and XGBoost. DEGnext was robust and effective in terms of transferring learned feature maps to facilitate classification of unseen datasets. Additionally, we validated that the predicted DEGs from DEGnext were mapped to significant Gene Ontology terms and pathways related to cancer.ConclusionsDEGnext can classify DEGs into UR and DR genes from RNA-seq cancer datasets with high performance. This type of analysis, using biologically relevant fine-tuning data, may aid in the exploration of potential biomarkers and can be adapted for other disease datasets.

Genome-wide transcriptome and translatome analyses reveal the role of protein extension and domestication in liver cancer oncogenesis

One gene could be transcribed to different RNA isoforms, and then produce various forms of protein sequences. This mechanism largely diversifies the cellular pool and allows natural selection to select from a wider range of substrates. In the cancer field, the isoform switches between tumor and normal tissues, such as the alternative splicing, stop codon read-through, or protein domestication, are significantly ignored by the traditional differential expression analyses. The intention of this work is to fill this gap. We collected public transcriptome and translatome data from ten patients with liver cancer, and performed genome-wide comparison on the stop codon read-through and protein domestication events. Both events diversify the proteome during long-term evolution. Surprisingly, we found that the tumor tissues globally have higher occurrence of stop codon read-through events as well as protein domestication events (translation signals of non-coding repetitive elements). These read-through and domestication events show limited overlapping across the ten patients, indicating the randomness of their occurrence and their deleterious nature. These tumor-specific events might have been purged by natural selection if they are not collected timely. Our work manifests the role of protein extension and domestication in liver cancer oncogenesis, adding new aspects to the cancer field.

Weighted gene co-expression network analysis unveils gene networks associated with the Fusarium head blight resistance in tetraploid wheat

BackgroundFusarium head blight (FHB) resistance in the durum wheat breeding gene pool is rarely reported. Triticum turgidum ssp. carthlicum line Blackbird is a tetraploid relative of durum wheat that offers partial FHB resistance. Resistance QTL were identified for the durum wheat cv. Strongfield × Blackbird population on chromosomes 1A, 2A, 2B, 3A, 6A, 6B and 7B in a previous study. The objective of this study was to identify the defense mechanisms underlying the resistance of Blackbird and report candidate regulator defense genes and single nucleotide polymorphism (SNP) markers within these genes for high-resolution mapping of resistance QTL reported for the durum wheat cv. Strongfield/Blackbird population.ResultsGene network analysis identified five networks significantly (P < 0.05) associated with the resistance to FHB spread (Type II FHB resistance) one of which showed significant correlation with both plant height and relative maturity traits. Two gene networks showed subtle differences between Fusarium graminearum-inoculated and mock-inoculated plants, supporting their involvement in constitutive defense. The candidate regulator genes have been implicated in various layers of plant defense including pathogen recognition (mainly Nucleotide-binding Leucine-rich Repeat proteins), signaling pathways including the abscisic acid and mitogen activated protein (MAP) kinase, and downstream defense genes activation including transcription factors (mostly with dual roles in defense and development), and cell death regulator and cell wall reinforcement genes. The expression of five candidate genes measured by quantitative real-time PCR was correlated with that of RNA-seq, corroborating the technical and analytical accuracy of RNA-sequencing.ConclusionsGene network analysis allowed identification of candidate regulator genes and genes associated with constitutive resistance, those that will not be detected using traditional differential expression analysis. This study also shed light on the association of developmental traits with FHB resistance and partially explained the co-localization of FHB resistance with plant height and maturity QTL reported in several previous studies. It also allowed the identification of candidate hub genes within the interval of three previously reported FHB resistance QTL for the Strongfield/Blackbird population and associated SNPs for future high resolution mapping studies.

DNLC: differential network local consistency analysis

BackgroundThe biological network is highly dynamic. Functional relations between genes can be activated or deactivated depending on the biological conditions. On the genome-scale network, subnetworks that gain or lose local expression consistency may shed light on the regulatory mechanisms related to the changing biological conditions, such as disease status or tissue developmental stages.ResultsIn this study, we develop a new method to select genes and modules on the existing biological network, in which local expression consistency changes significantly between clinical conditions. The method is called DNLC: Differential Network Local Consistency. In simulations, our algorithm detected artificially created local consistency changes effectively. We applied the method on two publicly available datasets, and the method detected novel genes and network modules that were biologically plausible.ConclusionsThe new method is effective in finding modules in which the gene expression consistency change between clinical conditions. It is a useful tool that complements traditional differential expression analyses to make discoveries from gene expression data. The R package is available at https://cran.r-project.org/web/packages/DNLC.

DECtp: Calling Differential Gene Expression Between Cancer and Normal Samples by Integrating Tumor Purity Information.

Identifying differentially expressed genes (DEGs) between tumor and normal samples is critical for studying tumorigenesis, and has been routinely applied to identify diagnostic, prognostic, and therapeutic biomarkers for many cancers. It is well-known that solid tumor tissue samples obtained from clinical settings are always mixtures of cancer and normal cells. However, the tumor purity information is more or less ignored in traditional differential expression analyses, which might decrease the power of differential gene identification or even bias the results. In this paper, we have developed a novel differential gene calling method called DECtp by integrating tumor purity information into a generalized least square procedure, followed by the Wald test. We compared DECtp with popular methods like t-test and limma on nine simulation datasets with different sample sizes and noise levels. DECtp achieved the highest area under curves (AUCs) for all the comparisons, suggesting that cancer purity information is critical for DEG calling between tumor and normal samples. In addition, we applied DECtp into cancer and normal samples of 14 tumor types collected from The Cancer Genome Atlas (TCGA) and compared the DEGs with those called by limma. As a result, DECtp achieved more sensitive, consistent, and biologically meaningful results and identified a few novel DEGs for further experimental validation.

Differential function analysis: identifying structure and activation variations in dysregulated pathways

Complex diseases are generally caused by the dysregulation of biological functions rather than individual molecules. Hence, a major challenge of the systematical study on complex diseases is how to capture the differentially regulated biological functions, e.g., pathways. The traditional differential expression analysis (DEA) usually considers the changed expression values of genes rather than functions. Meanwhile, the conventional function-based analysis (e.g., PEA: pathway enrichment analysis) mainly considers the varying activation of functions but disregards the structure change of genetic elements of functions. To achieve precision medicine against complex diseases, it is necessary to distinguish both the changes of functions and their elements from heterogeneous dysregulated pathways during the disease development and progression. In this work, in contrast to the traditional DEA, we developed a new computational framework, namely differential function analysis (DFA), to identify the changes of element-structure and expression-activation of biological functions, based on comparative non-negative matrix factorization (cNMF). To validate the effectiveness of our method, we tested DFA on various datasets, which shows that DFA is able to effectively recover the differential element-structure and differential activation-score of pre-set functional groups. In particular, the analysis of DFA on human gastric cancer dataset, not only capture the changed network-structure of pathways associated with gastric cancer, but also detect the differential activations of these pathways (i.e., significantly discriminating normal samples and disease samples), which is more effective than the state-of-the-art methods, such as GSVA and Pathifier. Totally, DFA is a general framework to capture the systematical changes of genes, networks and functions of complex diseases, which not only provides the new insight on the simultaneous alterations of pathway genes and pathway activations, but also opens a new way for the network-based functional analysis on heterogeneous diseases.

Differential Coexpression Analysis Reveals Extensive Rewiring of Arabidopsis Gene Coexpression in Response to Pseudomonas syringae Infection.

Plant defense responses to pathogens involve massive transcriptional reprogramming. Recently, differential coexpression analysis has been developed to study the rewiring of gene networks through microarray data, which is becoming an important complement to traditional differential expression analysis. Using time-series microarray data of Arabidopsis thaliana infected with Pseudomonas syringae, we analyzed Arabidopsis defense responses to P. syringae through differential coexpression analysis. Overall, we found that differential coexpression was a common phenomenon of plant immunity. Genes that were frequently involved in differential coexpression tend to be related to plant immune responses. Importantly, many of those genes have similar average expression levels between normal plant growth and pathogen infection but have different coexpression partners. By integrating the Arabidopsis regulatory network into our analysis, we identified several transcription factors that may be regulators of differential coexpression during plant immune responses. We also observed extensive differential coexpression between genes within the same metabolic pathways. Several metabolic pathways, such as photosynthesis light reactions, exhibited significant changes in expression correlation between normal growth and pathogen infection. Taken together, differential coexpression analysis provides a new strategy for analyzing transcriptional data related to plant defense responses and new insights into the understanding of plant-pathogen interactions.

Systematic pan-cancer analysis of tumour purity.

The tumour microenvironment is the non-cancerous cells present in and around a tumour, including mainly immune cells, but also fibroblasts and cells that comprise supporting blood vessels. These non-cancerous components of the tumour may play an important role in cancer biology. They also have a strong influence on the genomic analysis of tumour samples, and may alter the biological interpretation of results. Here we present a systematic analysis using different measurement modalities of tumour purity in >10,000 samples across 21 cancer types from the Cancer Genome Atlas. Patients are stratified according to clinical features in an attempt to detect clinical differences driven by purity levels. We demonstrate the confounding effect of tumour purity on correlating and clustering tumours with transcriptomics data. Finally, using a differential expression method that accounts for tumour purity, we find an immunotherapy gene signature in several cancer types that is not detected by traditional differential expression analyses.

DECODE: an integrated differential co-expression and differential expression analysis of gene expression data.

BackgroundBoth differential expression (DE) and differential co-expression (DC) analyses are appreciated as useful tools in understanding gene regulation related to complex diseases. The performance of integrating DE and DC, however, remains unexplored.ResultsIn this study, we proposed a novel analytical approach called DECODE (Differential Co-expression and Differential Expression) to integrate DC and DE analyses of gene expression data. DECODE allows one to study the combined features of DC and DE of each transcript between two conditions. By incorporating information of the dependency between DC and DE variables, two optimal thresholds for defining substantial change in expression and co-expression are systematically defined for each gene based on chi-square maximization. By using these thresholds, genes can be categorized into four groups with either high or low DC and DE characteristics. In this study, DECODE was applied to a large breast cancer microarray data set consisted of two thousand tumor samples. By identifying genes with high DE and high DC, we demonstrated that DECODE could improve the detection of some functional gene sets such as those related to immune system, metastasis, lipid and glucose metabolism. Further investigation on the identified genes and the associated functional pathways would provide an additional level of understanding of complex disease mechanism.ConclusionsBy complementing the recent DC and the traditional DE analyses, DECODE is a valuable methodology for investigating biological functions of genes exhibiting disease-associated DE and DC combined characteristics, which may not be easily revealed through DC or DE approach alone.DECODE is available at the Comprehensive R Archive Network (CRAN): http://cran.r-project.org/web/packages/decode/index.html.Electronic supplementary materialThe online version of this article (doi:10.1186/s12859-015-0582-4) contains supplementary material, which is available to authorized users.

Dynamic protein interaction modules in human hepatocellular carcinoma progression

BackgroundGene expression profiles have been frequently integrated with the human protein interactome to uncover functional modules under specific conditions like disease state. Beyond traditional differential expression analysis, differential co-expression analysis has emerged as a robust approach to reveal condition-specific network modules, with successful applications in a few human disease studies. Hepatocellular carcinoma (HCC), which is often interrelated with the Hepatitis C virus, typically develops through multiple stages. A comprehensive investigation of HCC progression-specific differential co-expression modules may advance our understanding of HCC's pathophysiological mechanisms.ResultsCompared with differentially expressed genes, differentially co-expressed genes were found more likely enriched with Hepatitis C virus binding proteins and cancer-mutated genes, and they were clustered more densely in the human reference protein interaction network. These observations indicated that a differential co-expression approach could outperform the standard differential expression network analysis in searching for disease-related modules. We then proposed a differential co-expression network approach to uncover network modules involved in HCC development. Specifically, we discovered subnetworks that enriched differentially co-expressed gene pairs in each HCC transition stage, and further resolved modules with coherent co-expression change patterns over all HCC developmental stages. Our identified network modules were enriched with HCC-related genes and implicated in cancer-related biological functions. In particular, APC and YWHAZ were highlighted as two most remarkable genes in the network modules, and their dynamic interaction partnership was resolved in HCC development.ConclusionsWe demonstrated that integration of differential co-expression with the protein interactome could outperform the traditional differential expression approach in discovering network modules of human diseases. In our application of this approach to HCC's gene expression data, we successfully identified subnetworks with marked differential co-expression in individual HCC stage transitions and network modules with coherent co-expression change patterns over all HCC developmental stages. Our results shed light on subtle HCC mechanisms, including temporal activation and dismissal of pivotal functions and dynamic interaction partnerships of key genes.

Coexpression of Nuclear Receptors and Histone Methylation Modifying Genes in the Testis: Implications for Endocrine Disruptor Modes of Action

BackgroundEndocrine disruptor chemicals elicit adverse health effects by perturbing nuclear receptor signalling systems. It has been speculated that these compounds may also perturb epigenetic mechanisms and thus contribute to the early origin of adult onset disease. We hypothesised that histone methylation may be a component of the epigenome that is susceptible to perturbation. We used coexpression analysis of publicly available data to investigate the combinatorial actions of nuclear receptors and genes involved in histone methylation in normal testis and when faced with endocrine disruptor compounds.Methodology/Principal FindingsThe expression patterns of a set of genes were profiled across testis tissue in human, rat and mouse, plus control and exposed samples from four toxicity experiments in the rat. Our results indicate that histone methylation events are a more general component of nuclear receptor mediated transcriptional regulation in the testis than previously appreciated. Coexpression patterns support the role of a gatekeeper mechanism involving the histone methylation modifiers Kdm1, Prdm2, and Ehmt1 and indicate that this mechanism is a common determinant of transcriptional integrity for genes critical to diverse physiological endpoints relevant to endocrine disruption. Coexpression patterns following exposure to vinclozolin and dibutyl phthalate suggest that coactivity of the demethylase Kdm1 in particular warrants further investigation in relation to endocrine disruptor mode of action.Conclusions/SignificanceThis study provides proof of concept that a bioinformatics approach that profiles genes related to a specific hypothesis across multiple biological settings can provide powerful insight into coregulatory activity that would be difficult to discern at an individual experiment level or by traditional differential expression analysis methods.

Making sense out of massive data by going beyond differential expression

With the rapid growth of publicly available high-throughput transcriptomic data, there is increasing recognition that large sets of such data can be mined to better understand disease states and mechanisms. Prior gene expression analyses, both large and small, have been dichotomous in nature, in which phenotypes are compared using clearly defined controls. Such approaches may require arbitrary decisions about what are considered "normal" phenotypes, and what each phenotype should be compared to. Instead, we adopt a holistic approach in which we characterize phenotypes in the context of a myriad of tissues and diseases. We introduce scalable methods that associate expression patterns to phenotypes in order both to assign phenotype labels to new expression samples and to select phenotypically meaningful gene signatures. By using a nonparametric statistical approach, we identify signatures that are more precise than those from existing approaches and accurately reveal biological processes that are hidden in case vs. control studies. Employing a comprehensive perspective on expression, we show how metastasized tumor samples localize in the vicinity of the primary site counterparts and are overenriched for those phenotype labels. We find that our approach provides insights into the biological processes that underlie differences between tissues and diseases beyond those identified by traditional differential expression analyses. Finally, we provide an online resource (http://concordia.csail.mit.edu) for mapping users' gene expression samples onto the expression landscape of tissue and disease.

Integrated genomics of susceptibility to alkylator-induced leukemia in mice

BackgroundTherapy-related acute myeloid leukemia (t-AML) is a secondary, generally incurable, malignancy attributable to chemotherapy exposure. Although there is a genetic component to t-AML susceptibility in mice, the relevant loci and the mechanism(s) by which they contribute to t-AML are largely unknown. An improved understanding of susceptibility factors and the biological processes in which they act may lead to the development of t-AML prevention strategies.ResultsIn this work we applied an integrated genomics strategy in inbred strains of mice to find novel factors that might contribute to susceptibility. We found that the pre-exposure transcriptional state of hematopoietic stem/progenitor cells predicts susceptibility status. More than 900 genes were differentially expressed between susceptible and resistant strains and were highly enriched in the apoptotic program, but it remained unclear which genes, if any, contribute directly to t-AML susceptibility. To address this issue, we integrated gene expression data with genetic information, including single nucleotide polymorphisms (SNPs) and DNA copy number variants (CNVs), to identify genetic networks underlying t-AML susceptibility. The 30 t-AML susceptibility networks we found are robust: they were validated in independent, previously published expression data, and different analytical methods converge on them. Further, the networks are enriched in genes involved in cell cycle and DNA repair (pathways not discovered in traditional differential expression analysis), suggesting that these processes contribute to t-AML susceptibility. Within these networks, the putative regulators (e.g., Parp2, Casp9, Polr1b) are the most likely to have a non-redundant role in the pathogenesis of t-AML. While identifying these networks, we found that current CNVR and SNP-based haplotype maps in mice represented distinct sources of genetic variation contributing to expression variation, implying that mapping studies utilizing either source alone will have reduced sensitivity.ConclusionThe identification and prioritization of genes and networks not previously implicated in t-AML generates novel hypotheses on the biology and treatment of this disease that will be the focus of future research.

DCGL: an R package for identifying differentially coexpressed genes and links from gene expression microarray data

Summary: Gene coexpression analysis was developed to explore gene interconnection at the expression level from a systems perspective, and differential coexpression analysis (DCEA), which examines the change in gene expression correlation between two conditions, was accordingly designed as a complementary technique to traditional differential expression analysis (DEA). Since there is a shortage of DCEA tools, we implemented in an R package ‘DCGL’ five DCEA methods for identification of differentially coexpressed genes and differentially coexpressed links, including three currently popular methods and two novel algorithms described in a companion paper. DCGL can serve as an easy-to-use tool to facilitate differential coexpression analyses.Contact: yyli@scbit.org and yxli@scbit.orgSupplementary information: Supplementary data are available at Bioinformatics online.