Phylogenetic Analysis of Gene Expression

In order to understand even the simplest cellular processes, we need to integrate proteomic, gene expression and other biomolecular data. To date, most computational approaches aimed at integrating proteomics and gene expression data use direct gene/protein correlation measures. However, due to post-transcriptional and translational regulations, the correspondence between the expression of a gene and its protein is complicated. We apply a multivariate statistical method, co-inertia analysis (CIA), to visualise gene and proteomic expression data stemming from the same biological samples. Principal components analysis or correspondence analysis can be used for data exploration on single datasets. CIA is then used to explore the relationships between two or more datasets. We further explore the data by projecting gene ontology (GO) information onto these plots to describe the cellular processes in action. We apply these techniques to gene expression and protein abundance data from studies of the human malarial parasite life cycle and the NCI-60 cancer cell lines. In each case, we visualise gene expression, protein abundance and GO classes in the same low dimensional projections and identify GO classes that are likely to be of biological importance.

Gene expression studies are increasingly used to provide valuable information on the diagnosis and prognosis of human cancers. Also, for in vitro and in vivo experimental cancer models gene expression studies are widely used. The complex algorithms of differential gene expression analyses require normalization of data against a reference or normalizer gene, or a set of such genes. For this purpose, mostly invariant housekeeping genes are used. Unfortunately, however, there are no consensus (housekeeping) genes that serve as reference or normalizer for different human cancers. In fact, scientists have employed a wide range of reference genes across different types of cancer for normalization of gene expression data. As a consequence, comparisons of these data and/or data harmonizations are difficult to perform and challenging. In addition, an inadequate choice for a reference gene may obscure genuine changes and/or result in erroneous gene expression data comparisons. In our effort to highlight the importance of selecting the most appropriate reference gene(s), we have screened the literature for gene expression studies published since the turn of the century on thirteen of the most prevalent human cancers worldwide. Based on the analysis of the data at hand, we firstly recommend that in each study the suitability of candidate reference gene(s) should carefully be evaluated in order to yield reliable differential gene expression data. Secondly, we recommend that a combination of PPIA and either GAPDH, ACTB, HPRT and TBP, or appropriate combinations of two or three of these genes, should be employed in future studies, to ensure that results from different studies on different human cancers can be harmonized. This approach will ultimately increase the depth of our understanding of gene expression signatures across human cancers.

BackgroundA complete understanding of the regulatory mechanisms of gene expression is the next important issue of genomics. Many bioinformaticians have developed methods and algorithms for predicting transcriptional regulatory mechanisms from sequence, gene expression, and binding data. However, most of these studies involved the use of yeast which has much simpler regulatory networks than human and has many genome wide binding data and gene expression data under diverse conditions. Studies of genome wide transcriptional networks of human genomes currently lag behind those of yeast.ResultsWe report herein a new method that combines gene expression data analysis with promoter analysis to infer transcriptional regulatory elements of human genes. The Z scores from the application of gene set analysis with gene sets of transcription factor binding sites (TFBSs) were successfully used to represent the activity of TFBSs in a given microarray data set. A significant correlation between the Z scores of gene sets of TFBSs and individual genes across multiple conditions permitted successful identification of many known human transcriptional regulatory elements of genes as well as the prediction of numerous putative TFBSs of many genes which will constitute a good starting point for further experiments. Using Z scores of gene sets of TFBSs produced better predictions than the use of mRNA levels of a transcription factor itself, suggesting that the Z scores of gene sets of TFBSs better represent diverse mechanisms for changing the activity of transcription factors in the cell. In addition, cis-regulatory modules, combinations of co-acting TFBSs, were readily identified by our analysis.ConclusionBy a strategic combination of gene set level analysis of gene expression data sets and promoter analysis, we were able to identify and predict many transcriptional regulatory elements of human genes. We conclude that this approach will aid in decoding some of the important transcriptional regulatory elements of human genes.

Current high-throughput gene expression experiments have a straightforward design of examining the gene expression of one group or condition relative to that of another. The data is typically analyzed as if they represent strictly intracellular events, and often treats genes as coming from a homogeneous population. Although intracellular events are crucial to nearly all biological processes, cell-cell interactions are often just as important, especially when gene expression data is generated from heterogeneous cell populations, such as from whole tissues. Cell-cell molecular interactions are generally lost in the available analytical procedures and as a result, are not examined experimentally, at least not accurately or with efficiency. Most importantly, this imposes major limitations when studying gene expression changes in multiple samples that interact with one another. In order to addresses the limitations of current techniques, we have developed a novel systems-based approach that expands the traditional analysis of gene expression in two stages. This includes a novel sequence-based meta-analytic tool, AbsIDconvert, that allows for conversion of annotated features using an interval tree for storing and querying absolute genomic coordinates for comparison of multi-scale macro-molecule identifiers across platforms and/or organisms. In addition, a systems-based heuristic algorithm is developed to find intercellular interactions between two sets of genes, potentially from different tissues by utilizing location information of each gene along with the information available in the secondary databases in the form of interactions, pathways and signaling. AbsIDconvert is shown to provide a high accuracy in identifier conversion as compared to other available methodologies (typically at an average rate of 84%) while maintaining a higher efficiency (O(n*log(n)). Our intercellular interaction approach and underlying visualization shows promise in allowing researchers to uncover novel signaling pathways in an intercellular fashion that to this point has not been possible.

Lack of reproducibility in gene expression studies is a serious issue being actively addressed by the biomedical research community. Besides established factors such as batch effects and incorrect sample annotations, we recently reported tissue heterogeneity, a consequence of unintended profiling of cells of other origins than the tissue of interest, as a source of variance. Although tissue heterogeneity exacerbates irreproducibility, its prevalence in gene expression data remains unknown. Here, we systematically analyse 2 667 publicly available gene expression datasets covering 76 576 samples. Using two independent data compendia and a reproducible, open-source software pipeline, we find a prevalence of tissue heterogeneity in gene expression data that affects between 1 and 40% of the samples, depending on the tissue type. We discover both cases of severe heterogeneity, which may be caused by mistakes in annotation or sample handling, and cases of moderate heterogeneity, which are likely caused by tissue infiltration or sample contamination. Our analysis establishes tissue heterogeneity as a widespread phenomenon in publicly available gene expression datasets, which constitutes an important source of variance that should not be ignored. Consequently, we advocate the application of quality-control methods such as BioQC to detect tissue heterogeneity prior to mining or analysing gene expression data.

Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Lung squamous cell carcinoma (LUSC) is a type of lung cancer with a dismal prognosis that lacks adequate therapies and actionable targets. This disease is characterized by a sequence of low- and high-grade preinvasive stages with increasing probability of malignant progression. Increasing our knowledge about the biology of these premalignant lesions (PMLs) is necessary to design new methods of early detection and prevention, and to identify the molecular processes that are key for malignant progression. To facilitate this research, we have designed XTABLE (Exploring Transcriptomes of Bronchial Lesions), an open-source application that integrates the most extensive transcriptomic databases of PMLs published so far. With this tool, users can stratify samples using multiple parameters and interrogate PML biology in multiple manners, such as two- and multiple-group comparisons, interrogation of genes of interests, and transcriptional signatures. Using XTABLE, we have carried out a comparative study of the potential role of chromosomal instability scores as biomarkers of PML progression and mapped the onset of the most relevant LUSC pathways to the sequence of LUSC developmental stages. XTABLE will critically facilitate new research for the identification of early detection biomarkers and acquire a better understanding of the LUSC precancerous stages. Editor's evaluation The authors have created a resource tool that is valuable in assessing precancerous lesions in the lung, which may serve as a tool for investigators working in this area, and as an example for additional similar resources. The accessibility of the tool is a concern but does not diminish the quality of the product. https://doi.org/10.7554/eLife.77507.sa0 Decision letter Reviews on Sciety eLife's review process Introduction Lung squamous cell carcinoma (LUSC) is a type of non-small cell lung cancer that accounts for 20–30% of all lung cancer cases (Chen et al., 2014; Sung et al., 2021; Bray et al., 2021). Despite being the second most frequent type of lung cancer (Torre et al., 2016), our knowledge regarding the biology of this disease as well as the therapeutic modalities to treat it remain far behind the most frequent type of lung cancer, lung adenocarcinoma (LUAD) (Hirsch et al., 2016; Khuder, 2001). LUAD genetics is dominated by mutations (that are often druggable) that activate the RTK/RAS pathway, including EGFR and KRAS mutations (The Cancer Genome Atlas Research Network, 2012; Jamal-Hanjani et al., 2017). However, the genetic landscape of LUSC is more complex, with multiple pathways altered in subsets of patients and a lack of actionable mutations (Campbell et al., 2016; The Cancer Genome Atlas Research Network, 2012; Kim et al., 2014), precluding the development of new therapies. Hence, the only pharmacological therapies available to treat LUSC patients are immune checkpoint inhibitor monotherapy or in combination with chemotherapy (Mok et al., 2019; Paz-Ares et al., 2018; Weinberg and Gadgeel, 2019). Furthermore, the National Lung Cancer Matrix Trial and The Lung Master Protocol, the largest personalized medicine trials in lung cancer, have not shown clear therapeutic benefits with targeted agents in LUSC (; Middleton et al., 2020; Redman et al., 2020). LUSC also has a worse prognosis than LUAD independent of stage at detection, with a 5-year overall survival (OS) of 6.2% for patients diagnosed with distant metastasis (9.5% for LUAD). However, patients diagnosed with localized disease are eligible for curative surgery and the 5-year OS is 50% (National Cancer Institute and DCCPS, 2018). Therefore, early detection is currently the most valuable tool to prevent deaths by LUSC as evidenced by several ongoing programmes of lung cancer early detection (Crosbie et al., 2019a; Crosbie et al., 2019b; Aberle et al., 2011). These initiatives make use of low-dose CT scans in high-risk populations, but in spite of the frequent detection of localized lung cancer eligible for resection with curative intent, 40% of early diagnosed patients die within 5 years. LUSC progresses through a series of premalignant stages characterized by alterations of the normal bronchial epithelium (Figure 1; Ishizumi et al., 2010; Kadara et al., 2016; Thakrar et al., 2017; Pennycuick et al., 2020). These endobronchial premalignant lesions (PMLs) are classified as low-grade (squamous metaplasia, mild and moderate dysplasia) and high-grade (severe dysplasia, and carcinomas in situ [CIS]) (Figure 1). However, not all PMLs progress to LUSC. Although obvious differences exist between the multiple studies published on the topic, high-grade PMLs have a higher risk of progression than low grade, and high levels of chromosomal instability (CIN) are also predictive of progressive PMLs (Thakrar et al., 2017; van Boerdonk et al., 2014; Teixeira et al., 2019; Merrick et al., 2016). This potential role of CIN as biomarker of PMLs progression has been observed in low- and high-grade PMLs (van Boerdonk et al., 2014; Teixeira et al., 2019). These reports showed that high levels of copy number variations was the best predictor of progressive PMLs (van Boerdonk et al., 2014; Teixeira et al., 2019) and that immune response is the most likely cause of regression as these lesions contained higher levels of immune infiltration (Pennycuick et al., 2020; Beane et al., 2019). These results are supported by transcriptomic analysis of PMLs that indicate immune evasion in the transition to invasiveness (Mascaux et al., 2019). Figure 1 Download asset Open asset Developmental stages of lung squamous cell carcinoma (LUSC) premalignant lesions (PMLs) with representative histological images for each stage (haematoxilyn-eosin) and a summary of the four studies included in XTABLE (Exploring Transcriptomes of Bronchial Lesions). PMLs are typically classified as normal epithelium (including hyperplasia), low-grade and high-grade. Two studies (Mascaux et al., 2019, and Beane et al., 2019) carried out gene expression analysis of multiple developmental stages, whereas Merrick et al., 2018, and Teixeira et al., 2019, focused on dysplasias (blue boxes) and carcinomas in situ (CIS) (pink boxes), respectively. The most relevant findings of each article are summarized in the figure. Error bars=50 µm. The detection of PMLs cannot be carried out by routine patient imaging techniques such as CT or PET scans as the morphological change that they cause in the airway does not result in radiological contrast. Alternatively, ablation of high-grade PMLs detected by autofluorescence bronchoscopy using minimally invasive endoscopic procedures in high-risk populations is an innovative and interesting strategy to prevent LUSC (Guibert et al., 2016). Nonetheless autofluorescence bronchoscopy is an expensive and complex technique of limited use in large screening programmes. Therefore, simpler, more cost effective, and scalable methods of high-grade PML detection are needed to prevent deaths by LUSC. Improving the detection of PMLs requires a better understanding of their biology and the validation of adequate biomarkers of progressive lesions that can be translated into new technologies for large screening initiatives. Cell surface proteins, metabolites, nasal-based biomarkers, blood and sputum/bronchoalveolar lavage biomarkers are examples of biomarkers that can be used to improve and/or complement current diagnostic techniques (such as CT and PET scans) or develop new ones. A user-friendly application to interrogate gene expression from studies focusing on precancerous LUSC stages will enhance the advances in PML biology. However, such application does not exist. Recently, scientific interest in the biology of preinvasive LUSC stages has motivated the publication of several articles characterizing PML transcriptomes from various perspectives (Figure 1). Two reports published by Beane et al., 2019, and Mascaux et al., 2019, showed stage-by-stage gene expression analyses of PMLs. These studies provided the most detailed transcriptomic characterization of all PML stages so far and identified changes in the immune microenvironment associated with invasive transformation. Additionally, longitudinal studies by Beane et al., 2019, Merrick et al., 2018, and Teixeira et al., 2019, contained samples with known progression potential. Beane et al., 2019, identified molecular subtypes with specific biological traits whereas articles published by Merrick et al., 2018, and Teixeira et al., 2019 focused on gene expression of specific preinvasive stages (dysplasias and CIS, respectively) with the objective of identifying predictors of PML progression and the biomolecular processes involved (Figure 1). These studies provide a valuable source of gene expression data to identify candidate biomarkers for the detection of high-risk PMLs and/or early stage LUSC as well as to investigate the biology of premalignant LUSC progression. Transparent and straightforward accessibility to transcriptomic databases is a key requirement for the open science philosophy. Applications that integrate multiple databases focusing on the same biological and clinical problem, such as camcAPP (Dunning et al., 2017), allow cross-comparisons between independent studies, strengthen the robustness of results obtained, and allow the selection of high-confidence data. In this report, we provide an overall description of XTABLE (Exploring Transcriptomes of Bronchial Lesions) and provide examples of its functions. XTABLE is a new open-source application that will enable scientists to interrogate currently four LUSC PML transcriptomic datasets in a versatile manner that can be adapted to the needs of each researcher. Specifically, LUSC prevention and diagnosis are the main areas that can benefit the most from XTABLE, but its versatility and multiple functions lend themselves to the exploration of a variety of research questions. Without XTABLE, researchers would have to put together all the packages, data processing steps themselves for each analysis they wished to run. In this report, we provide an overall explanation of all the functions of XTABLE as well as a detailed description of the most important analysis modules, including two-group comparisons, gene-of-interest analyses, and interrogation of transcriptional signatures. Additionally, we explored the use of CIN-related signatures as a biomarker of progressive PMLs and mapped the onset of the most important LUSC pathways to its developmental stages. Results Description of the studies included in XTABLE Four datasets originating from four independent PML transcriptomic studies have been included in the XTABLE application (Figure 1, Table 1). Two datasets, GSE109743 (Beane et al., 2019) and GSE33479 (Mascaux et al., 2019), provide gene expression data of the developmental LUSC stages (with and without progression status information, respectively), whereas the remaining studies, GSE114489 (Merrick et al., 2018) and GSE108124 (Teixeira et al., 2019), focus on analysing specific PML stages (dysplasias and CIS, respectively) that have been followed up to establish their progressive, persistent, or regressive potential. Table 1 Description of the four cohorts included in XTABLE (Exploring Transcriptomes of Bronchial Lesions). GEO accessionPMIDStagesProgression status knownNumber of samplesSample typeTranscriptomeGSE3347931243362MultipleNo122Whole biopsiesMicroarrayGSE10974331015447Multiple*Yes448Discovery cohort (197 Bx, 91 Br)Validation cohort (111Bx, 49 Br)Whole biopsies and brushingsRNAseqGSE11448929997230Dysplasias normalYes63Whole biopsiesMicroarrayGSE10812430664780CISYes33MicrodissectedMicroarray Bx: biopsies; Br: brushings. * This cohort includes neither carcinomas in situ (CIS) nor invasive carcinomas. Dataset GSE33479 (Mascaux et al., 2019) comprises expression microarray analyses of 122 endobronchial biopsies of unknown progression status. Samples were obtained following autofluorescence bronchoscopy and as no enrichment or microdissection of the biopsy epithelium was carried out, variable levels of stromal component are present in samples. Although this results in dilution of epithelial signals, it has the advantage of providing information about the microenvironment as well as insights into the infiltrated immune cells. Dataset GSE109743 (Beane et al., 2019) consists of a transcriptomic analysis of whole PML biopsies with no purification of the epithelial compartment as well as bronchial brushings obtained from adjacent normal regions of the bronchial mucosa. Unlike GSE33479, GSE109743 used RNAseq and more importantly, includes the progression status for some lesions established by serial biopsies. Samples were classified as ‘normal-stable’ when they changed between normal, hyperplasia, and metaplasia, ‘regressive’ when they regress from dysplasia to a less severe dysplastic grade, or from dysplasia to normal/hyperplasia/metaplasia. Remaining samples were considered persistent/progressive. One advantage of this dataset is the large number of samples divided into a discovery and a validation cohort (Figure 1, Table 1). However, the representation of each PML stage is not homogeneous and CIS samples are not included. Datasets GSE114489 (Merrick et al., 2018) and GSE108124 (Teixeira et al., 2019) constitute a different type of study. Both make use of sequential biopsies to classify lesions according to their progression potential, but they differ in the stages included, the classification of progression status and sample processing. In GSE114489, the authors collected 63 baseline bronchial biopsies (with corresponding follow-up biopsies) and classified samples in four groups: 23 persistent dysplasias (dysplastic lesions with the same or higher severity scores in follow-up biopsies), 15 regressive dysplasias (dysplasias progressing to lower severity scores), 9 progressive non-dysplasias (biopsies with normal or hyperplastic morphologies that progress to more severe morphologies), and 16 stable non-dysplasias (normal or hyperplastic pathology that remains stable in follow-up biopsies). Microarray analysis of gene expression was performed on whole biopsies. The study published by Teixeira et al., 2019 (GSE108124) also makes use of follow-up biopsies to classify the progression potential of PMLs, but unlike GSE114489, GSE108124 focuses on CIS and progression is defined as the transition to invasive carcinomas in follow-up biopsies. RNA for gene expression microarray analysis was extracted from microdissected FFPE samples to enrich the epithelial component. XTABLE access, interface, and functions XTABLE download can be carried out from https://gitlab.com/cruk-mi/xtable (copy archived at Roberts, 2022) following the instructions in the Materials and methods section and Video 1. XTABLE has been designed using the shiny app interface (see Materials and methods section) and its functions have been divided into 11 interrelated tabs that contain a specific function to interrogate each dataset separately (Figure 2A). Video 1 Download asset This video cannot be played in place because your browser does support HTML5 video. You may still download the video for offline viewing. Download as MPEG-4 Download as WebM Download as Ogg Step-by-step instructions to install XTABLE (Exploring Transcriptomes of Bronchial Lesions) using RStudio. Figure 2 with 3 supplements see all Download asset Open asset Overall organization of XTABLE (Exploring Transcriptomes of Bronchial Lesions) functions and use of the DEG function. (A) Organization of all the functions in the XTABLE interface. The functions are interrelated and completing certain analyses requires the use of several functions. For instance, the GSEA and PA functions operate with gene lists obtained with the DEG function. (B) Workflow to obtain differentially expressed genes between two groups using the DEG function. The example shows groups of samples arranged by developmental stage to compare low-grade and high-grade premalignant lesions (PMLs) in the GSE33479 cohort. (C and D) Workflow to obtain differentially expressed genes between two groups using the DEG function. The two groups have been arranged by progression status using in the GSE114489 and GSE108124 cohorts, respectively. Introduction Detailed description of the four studies, analyses performed in each tab, output formats, links to the four articles used in the application, bioinformatic packages used in the different functions and additional references. Dataset tab Explanation of the differences in methodology and progression status definitions between the four studies. Load tab In this tab, the user selects the database (GSE114489, GSE108124, GSE109743, or GSE33479) for subsequent interrogation. Due to specific differences between the four studies, only one dataset can be interrogated at a time. Since dataset GSE109743 contains a discovery and a validation cohort, as well as biopsies and brushings, preselection of the cohort and type of sample has to be carried out (Figure 2—figure supplement 1). CIN-score tab CIN has been identified as a good predictor of PML progression. Except for Teixeira et al., 2019, the other three studies do not provide genomic analyses that provide an estimate of the level of chromosomal alterations in the samples. Several gene expression signatures that correlate with CIN (CIN-scores) have been described in the literature (Teixeira et al., 2019; Carter et al., 2006). This CIN-score function returns a list of several CIN-scores (CIN70, CIN25, and CIN5, depending on the number of genes included in the signature) for all samples included in the study and a graph depicting CIN-scores in different sample types classified according to the sample classification defined in the study. A line marking a selected CIN-score threshold can be added for visualization purpose and to help determine appropriate thresholds for additional CIN-related analyses (Figure 2—figure supplement 2). DEG tab Function to identify differentially expressed genes in comparisons of two groups of samples determined by the user. This function allows the selection of p-value and fold-change cutoffs for the analysis. Additionally, AUC is calculated for each gene to provide further confidence to the differentially expressed gene results and additional gene IDs are added at this stage that have been sourced from two separate datasets to maximize identification and discrepancies manually reviewed to improve downstream enrichment and pathway analyses. GSEA and PA In these two tabs, the user can carry out gene set and pathway enrichment analyses using multiple tools (goseq/ideal, fgsea/MSigDB, enrichR, gage/gageData, kegga/pathview, ReactomePA, Progeny, and Dorothea). This function operates with the list of differentially expressed genes obtained with the DEG function. Signature This function returns the gene expression values of a user-defined gene list. Lists can be manually entered or uploaded from a .csv file. Deconvolution Estimation of immune and stromal component in samples from gene expression data using the ESTIMATE tool. ROC This function returns receiver operating characteristic (ROC) curves in a user-defined comparison of two groups of samples stratified by the expression of a gene or by CIN-scores. Heatmap Returns a gene expression heatmap for genes selected by variance, user-defined gene signatures, and differentially expressed genes in the DEG tab. PCA Principal component analysis (PCA) of all the samples included in each study. Several sample characteristics can be highlighted in the PCA plot including CIN signatures, progression status, and PML analysis of expression of genes selected by the user. Several for sample are expression analysis The discovery of candidate biomarkers for the detection of PMLs at high risk of malignant progression and the interrogation of PML biology on the comparison of gene expression between lesions with known progressive or regressive potential. The four databases included in XTABLE contain different types of information which users can interrogate these To facilitate the interrogation of the four transcriptomic databases in a manner that allows the versatile of we have designed a in XTABLE DEG (Figure that returns the differentially expressed genes in two user-defined groups of samples stratified by PML stage and by known progression status GSE114489, and or by CIN-score In XTABLE, we have included three et al., and (Teixeira et al., 2019). and have been in the with the the number of genes for interrogation. is from the used by Teixeira et al., 2019, and is to a good with progression in CIS groups by stage using DEG allows the comparison of two PML stages or the of stages into two For instance, to compare the expression between low-grade and high-grade PMLs, we can two one including metaplasia, low and moderate dysplasia and one including severe dysplasia and CIS using cohort GSE33479 (Figure selection of the cohort and up the groups (Figure the application returns a list of differentially expressed genes with associated information (Figure including AUC by ROC analysis. ROC curves associated with each gene in a two-group for biomarker can be using the ROC tab (Figure 2—figure supplement of progressive and regressive samples can also be carried out by the in studies that provide this information (Figure and To carry out two-group comparison by CIN signatures, the CIN-score tab a visualization of the three CIN-scores for all samples to the selection of a CIN-score threshold for sample (Figure A .csv the CIN-scores of all samples can be The selected CIN threshold can be used in the DEG tab to stratify and low samples and a list of differentially expressed genes (Figure Figure 3 Download asset Open asset expression analysis between two groups of samples classified according to a chromosomal instability The CIN-score function allows the visualization of CIN-scores for all samples in a study. A CIN-score threshold selected by the user can be on the graph The CIN-score threshold selected by the user can be used in the DEG tab to the two-group the developmental stages of LUSC as described in Mascaux et al., 2019 CIN25, and can be used in the DEG tab. 1, and 2 and 3 and and the and largest within The list of differentially expressed genes obtained in the DEG tab can be used in the GSEA and PA tabs to carry out gene set enrichment and pathway analyses. The GSEA tab allows the user to the gene set enrichment tool fgsea/MSigDB, enrichR, and p-value and the gene to (Figure For instance, using the function to one of the three and molecular to for analysis (Figure with the tool, the user can the gene set of interest (Figure The PA tab operates in a similar manner using four pathway analysis tools ReactomePA, Progeny, and (Figure supplement 1). Figure with 1 supplement see all Download asset Open asset set enrichment analyses in a list differentially expressed genes using the GSEA tab. (A) set enrichment analysis using the tool of a list of differentially expressed genes obtained in the DEG tab. One of the three main can be selected for analysis at a time. selection of a XTABLE (Exploring Transcriptomes of Bronchial Lesions) returns a list of with associated (B) set enrichment analysis using the tool. This tool allows the selection of included in and returns a list of signatures with associated The example shows the selection of the in the analysis and user-defined transcriptional signatures can investigate their gene or of genes of interest in the To facilitate this type of analyses, we have included the and Signature functions in In the tab, the user can the expression of one gene of This tab is divided in three The tool returns a list of the multiple to the same gene expression values for the gene of interest in all samples (Figure The function returns the differentially expressed gene analysis results including fold-change and values for a gene in all groups of samples with a (Figure and on the study. The function to the function but allows of up to four groups of samples and the for analysis can be determined by the user. The example in Figure shows the of expression in four groups of samples that normal, low-grade PMLs, high-grade PMLs, and invasive carcinomas (Figure The and are the result with the comparison with the (normal normal and as the (Figure Figure 5 Download asset Open asset XTABLE (Exploring Transcriptomes of Bronchial Lesions) functions to analyses on genes and user defined gene signatures (A) The function the the expression values for a gene of interest in all samples. (B) The tool the expression of a gene of interest in groups of samples with a In the the function the expression of in all stages with the normal in The tool the of samples in up to four groups and comparison with a determined by the user. The example shows the analysis of expression in four groups of samples from the GSE33479 cohort and invasive The is set as for analysis. of the use of the Heatmap tab to interrogate to the expression of gene in premalignant lesions can be defined by the user in the and are shown using the stage classification and entered manually or from a .csv file. Alternatively, the heatmap can be from a list of differentially expressed genes from the DEG tab or a selected number of genes by The three can be selected in the In the example the heatmap shows all microarray associated to each gene calculated using To facilitate the interrogation of biological processes PML XTABLE also allows the interrogation of transcriptomic signatures using multiple functions. The Signature tab returns a list of expression values and a graph with the scores of expression for a gene set determined by the user in all samples of the selected study. The Heatmap tab returns a heatmap that the expression values in all samples in a selected study. The gene set shown in the heatmap can be selected using three from the With the the user can manually or a list of The example in Figure shows a heatmap from the GSE33479 dataset with transcriptional of an important LUSC The selects the list of genes differentially expressed in the DEG tab. the selects genes using a user-defined number of genes with the of the between CIN signatures and progression potential Two studies have highlighted the potential role of CIN as predictor of low-grade (van Boerdonk et al., and high-grade (Teixeira et al., 2019) PMLs progression in LUSC. XTABLE can be used to

Chronic alcohol consumption is associated with fatty liver disease in mammals. The object of this study was to gain an understanding of dysregulated lipid metabolism in alcohol-fed C57BL/6 mice using a targeted lipidomic approach. Liquid chromatography tandem mass spectrometry was used to analyze several lipid classes, including free fatty acids, fatty acyl-CoAs, fatty acid ethyl esters, sphingolipids, ceramides, and endocannabinoids, in plasma and liver samples from control and alcohol-fed mice. The interpretation of lipidomic data was augmented by gene expression analyses for important metabolic enzymes in the lipid pathways studied. Alcohol feeding was associated with i) increased hepatic free fatty acid levels and decreased fatty acyl-CoA levels associated with decreased mitochondrial fatty acid oxidation and decreased fatty acyl-CoA synthesis, respectively; ii) increased hepatic ceramide levels associated with higher levels of the precursor molecules sphingosine and sphinganine; and iii) increased hepatic levels of the endocannabinoid anandamide associated with decreased expression of its catabolic enzyme fatty acid amide hydrolase. The unique combination of lipidomic and gene expression analyses allows for a better mechanistic understanding of dysregulated lipid metabolism in the development of alcoholic fatty liver disease.

The real-time system of the JT-60 has a cascade structure with the major and the minor control loops. The multi-variable feedback control is adopted in the major control loop in order to adjust non-linear effects which will appear in the high-/beta/region. By applying the concept of 'phase shift control' to the control program of the major control loop, a flexible tokamak discharge control is expected to be performed with taking account of machine protection.

The emergence of large-scale multi-omics data warrants method development for data integration. Genomic studies from cancer patients have identified epigenetic and genetic regulators – such as methylation marks, somatic mutations, and somatic copy number alterations (SCNAs), among others – as predictive features of cancer outcome. However, identification of “driver genes” associated with a given alteration remains a challenge. To this end, we developed a computational tool, iEDGE, to model cis and trans effects of (epi-)DNA alterations and identify potential cis driver genes, where cis and trans genes denote those genes falling within and outside the genomic boundaries of a given (epi-)genetic alteration, respectively. iEDGE first identifies the cis and trans gene expression signatures associated with the presence/absence of a particular epi-DNA alteration across samples. It then applies tests of statistical mediation to determine the cis genes predictive of the trans gene expression. Finally, cis and trans effects are annotated by pathway enrichment analysis to gain insights into the underlying regulatory networks. We used iEDGE to perform integrative analysis of SCNAs and gene expression data from breast cancer and 18 additional cancer types included in The Cancer Genome Atlas (TCGA). Notably, cis gene drivers identified by iEDGE were found to be significantly enriched for known driver genes from multiple compendia of validated oncogenes and tumor suppressors, suggesting that the remainder are of equal importance. Furthermore, predicted drivers were enriched for functionally relevant cancer genes with amplification-driven dependencies, which are of potential prognostic and therapeutic value. All the analyses results are accessible at https://montilab.bu.edu/iEDGE. In summary, integrative analysis of SCNAs and gene expression using iEDGE successfully identified known cancer driver genes and putative cancer therapeutic targets across 19 cancer types in the TCGA. The proposed method can easily be applied to the integration of gene expression profiles with other epi-DNA assays in a variety of disease contexts.

BackgroundIn the recent years, there has been a rise in gene expression profiling reports. Unfortunately, it has not been possible to make maximum use of available gene expression data. Many databases and programs can be used to derive the possible expression patterns of mammalian genes, based on existing data. However, these available resources have limitations. For example, it is not possible to obtain a list of genes that are expressed in certain conditions. To overcome such limitations, we have taken up a new strategy to predict gene expression patterns using available information, for one tissue at a time.ResultsThe first step of this approach involved manual collection of maximum data derived from large-scale (genome-wide) gene expression studies, pertaining to mammalian testis. These data have been compiled into a Mammalian Gene Expression Testis-database (MGEx-Tdb). This process resulted in a richer collection of gene expression data compared to other databases/resources, for multiple testicular conditions. The gene-lists collected this way in turn were exploited to derive a 'consensus' expression status for each gene, across studies. The expression information obtained from the newly developed database mostly agreed with results from multiple small-scale studies on selected genes. A comparative analysis showed that MGEx-Tdb can retrieve the gene expression information more efficiently than other commonly used databases. It has the ability to provide a clear expression status (transcribed or dormant) for most genes, in the testis tissue, under several specific physiological/experimental conditions and/or cell-types.ConclusionsManual compilation of gene expression data, which can be a painstaking process, followed by a consensus expression status determination for specific locations and conditions, can be a reliable way of making use of the existing data to predict gene expression patterns. MGEx-Tdb provides expression information for 14 different combinations of specific locations and conditions in humans (25,158 genes), 79 in mice (22,919 genes) and 23 in rats (14,108 genes). It is also the first system that can predict expression of genes with a 'reliability-score', which is calculated based on the extent of agreements and contradictions across gene-sets/studies. This new platform is publicly available at the following web address: http://resource.ibab.ac.in/MGEx-Tdb/

An Approach to Virtual Karyotyping for Unbalanced Rearrangements Based on Gene Expression Data.

Prediction of hepatotoxicity in humans remains an unresolved challenge. Recently, an in vitro/in silico method was established to predict blood concentrations of test compounds with an increased risk of causing human hepatotoxicity. In the present study, we addressed the question whether gene expression data can improve the quality of hepatotoxicity prediction compared to cytotoxicity analysis alone. A particular challenge is that high-dimensional gene expression data must be sum­marized into variables that allow the determination of the lowest test compound concentration that causes altered gene expression. To address this challenge, we analyzed 60 hepatotoxic and non-hepatotoxic substances in a concentration-dependent manner for cytotoxicity and expression of 3,524 probes previously reported to be influenced by hepatotoxicants. The toxicity separation index (TSI) was applied to quantify how well specific summary variables of gene expression can differen­tiate between the set of hepatotoxic and non-hepatotoxic substances. The best TSI was obtained when the lowest concentration of a test compound that led to differential expression of two genes when compared to vehicle controls was considered positive. Furthermore, the best gene expression-based summary variable was superior to cytotoxicity-based variables alone, and the combination of the best summary variables of gene expression and cytotoxicity data further improved the TSI compared to each category alone. In conclusion, the method used to derive summary variables of gene expression is critical, and the best summary variables improve the prediction of hepatotoxic substances in relation to oral doses and blood concentrations in humans. Plain language summary Liver injury is a relevant side effect of many drugs. It represents the most common cause of acute liver failure in Western countries as well as the most common reason for late-stage drug development failure or withdrawal of drugs from the market. We show that a genome-wide gene expression analysis of cultured human hepatocytes exposed to drugs can identify hepatotoxic and non-hepato­toxic substances. To achieve this, we have developed a technique to summarize all gene expression variables to determine the lowest concentration at which a test substance begins to influence gene expression. We find that the cell test must include the highest concentration of the drug that is intended to be reached in the blood of patients at therapeutic doses. This approach can reduce the number of animals needed in pre-clinical safety studies as liver-damaging drug candidates would be detected and eliminated prior to studies in rodents.

The egg yolk precursor protein vitellogenin is widely used as a biomarker of estrogen exposure in male fish. However, standardized methodology is lacking and little is known regarding the reproducibility of results among laboratories using different equipment, reagents, protocols, and data analysis programs. To address this data gap we tested the reproducibility across laboratories to evaluate vitellogenin gene (vtg) expression and assessed the value of using a freely available software data analysis program. Samples collected from studies of male fathead minnows (Pimephales promelas) exposed to 17α-ethinylestradiol (EE2) and minnows exposed to processed wastewater effluent were evaluated for vtg expression in 4 laboratories. Our results indicate reasonable consistency among laboratories if the free software for expression analysis LinRegPCR is used, with 3 of 4 laboratories detecting vtg in fish exposed to 5 ng/L EE2 (n = 5). All 4 laboratories detected significantly increased vtg levels in 15 male fish exposed to wastewater effluent compared with 15 male fish held in a control stream. Finally, we were able to determine that the source of high interlaboratory variability from complementary deoxyribonucleic acid (cDNA) to quantitative polymerase chain reaction (qPCR) analyses was the expression analysis software unique to each real-time qPCR machine. We successfully eliminated the interlaboratory variability by reanalyzing raw fluorescence data with independent freeware, which yielded cycle thresholds and polymerase chain reaction (PCR) efficiencies that calculated results independently of proprietary software. Our results suggest that laboratories engaged in monitoring programs should validate their PCR protocols and analyze their gene expression data following the guidelines established in the present study for all gene expression biomarkers. Environ Toxicol Chem 2017;36:3102-3107. Published 2017 Wiley Periodicals Inc. on behalf of SETAC. This article is a US government work and, as such, is in the public domain in the United States of America.

BackgroundThe quantification of gene expression in tissue samples requires the use of reference genes to normalise transcript numbers between different samples. Reference gene stability may vary between different tissues, and between the same tissue in different disease states. We evaluated the stability of 9 reference genes commonly used in human gene expression studies. Real-time reverse transcription PCR and a mathematical algorithm were used to establish which reference genes were most stably expressed in normal and diseased canine articular tissues and two canine cell lines stimulated with lipolysaccaride (LPS).ResultsThe optimal reference genes for comparing gene expression data between normal and diseased infrapatella fat pad were RPL13A and YWHAZ (M = 0.56). The ideal reference genes for comparing normal and osteoarthritic (OA) cartilage were RPL13A and SDHA (M = 0.57). The best reference genes for comparing normal and ruptured canine cranial cruciate ligament were B2M and TBP (M = 0.59). The best reference genes for normalising gene expression data from normal and LPS stimulated cell lines were SDHA and YWHAZ (K6) or SDHA and HMBS (DH82), which had expression stability (M) values of 0.05 (K6) and 0.07 (DH82) respectively. The number of reference genes required to reduce pairwise variation (V) to <0.20 was 4 for cell lines, 5 for cartilage, 7 for cranial cruciate ligament and 8 for fat tissue. Reference gene stability was not related to the level of gene expression.ConclusionThe reference genes demonstrating the most stable expression within each different canine articular tissue were identified, but no single reference gene was identified as having stable expression in all different tissue types. This study underlines the necessity to select reference genes on the basis of tissue and disease specific expression profile evaluation and highlights the requirement for the identification of new reference genes with greater expression stability for use in canine articular tissue gene expression studies.

This work presents DeepReGraph, a novel method for co-clustering genes and cis-regulatory elements (CREs) into candidate regulatory networks. Gene expression data, as well as data from three CRE activity markers from a publicly available dataset of mouse fetal heart tissue, were used for DeepReGraph concept proofing. In this study we used open chromatin accessibility from ATAC-seq experiments, as well as H3K27ac and H3K27me3 histone marks as CREs activity markers. However, this method can be executed with other sets of markers. We modelled all data sources as a heterogeneous graph and adapted a state-of-the-art representation learning algorithm to produce a low-dimensional and easy-to-cluster embedding of genes and CREs. Deep graph auto-encoders and an adaptive-sparsity generative model are the algorithmic core of DeepReGraph. The main contribution of our work is the design of proper combination rules for the heterogeneous gene expression and CRE activity data and the computational encoding of well-known gene expression regulatory mechanisms into a suitable objective function for graph embedding. We showed that the co-clusters of genes and CREs in the final embedding shed light on developmental regulatory mechanisms in mouse fetal-heart tissue. Such clustering could not be achieved by using only gene expression data. Function enrichment analysis proves that the genes in the co-clusters are involved in distinct biological processes. The enriched transcription factor binding sites in CREs prioritize the candidate transcript factors which drive the temporal changes in gene expression. Consequently, we conclude that DeepReGraph could foster hypothesis-driven tissue development research from high-throughput expression and epigenomic data. Full source code and data are available on the DeepReGraph GitHub project.