Super-delta: a new differential gene expression analysis procedure with robust data normalization

Differential gene expression analysis is one of the significant efforts in single cell RNA sequencing (scRNAseq) analysis to discover the specific changes in expression levels of individual cell types. Since scRNAseq exhibits multimodality, large amounts of zero counts, and sparsity, it is different from the traditional bulk RNA sequencing (RNAseq) data. The new challenges of scRNAseq data promote the development of new methods for identifying differentially expressed (DE) genes. In this study, we proposed a new method, SigEMD, that combines a logistic regression model and a nonparametric method based on Earth Mover's Distance, to precisely and efficiently identify DE genes in scRNAseq data. The regression model is used to reduce the impact of large amounts of zero counts, and the nonparametric method is used to improve the sensitivity of detecting DE genes from multimodal scRNAseq data. By additionally employing gene interaction network information to adjust the final states of DE genes, we further reduce the false positives of calling DE genes. We used simulated data and real data to evaluate the detection accuracy of the proposed method and to compare its performance with those of other differential expression analysis methods. Results indicate that the proposed method has an overall powerful performance in terms of precision in detection, sensitivity, and specificity.

BackgroundDifferential expression and feature selection analyses are essential steps for the development of accurate diagnostic/prognostic classifiers of complicated human diseases using transcriptomics data. These steps are particularly challenging due to the curse of dimensionality and the presence of technical and biological noise. A promising strategy for overcoming these challenges is the incorporation of pre-existing transcriptomics data in the identification of differentially expressed (DE) genes. This approach has the potential to improve the quality of selected genes, increase classification performance, and enhance biological interpretability. While a number of methods have been developed that use pre-existing data for differential expression analysis, existing methods do not leverage the identities of experimental conditions to create a robust metric for identifying DE genes.ResultsIn this study, we propose a novel differential expression and feature selection method—GEOlimma—which combines pre-existing microarray data from the Gene Expression Omnibus (GEO) with the widely-applied Limma method for differential expression analysis. We first quantify differential gene expression across 2481 pairwise comparisons from 602 curated GEO Datasets, and we convert differential expression frequencies to DE prior probabilities. Genes with high DE prior probabilities show enrichment in cell growth and death, signal transduction, and cancer-related biological pathways, while genes with low prior probabilities were enriched in sensory system pathways. We then applied GEOlimma to four differential expression comparisons within two human disease datasets and performed differential expression, feature selection, and supervised classification analyses. Our results suggest that use of GEOlimma provides greater experimental power to detect DE genes compared to Limma, due to its increased effective sample size. Furthermore, in a supervised classification analysis using GEOlimma as a feature selection method, we observed similar or better classification performance than Limma given small, noisy subsets of an asthma dataset.ConclusionsOur results demonstrate that GEOlimma is a more effective method for differential gene expression and feature selection analyses compared to the standard Limma method. Due to its focus on gene-level differential expression, GEOlimma also has the potential to be applied to other high-throughput biological datasets.

Racial/ethnic disparity in cancer refers to the disproportionate incidence of and/or mortality from various cancers among population groups. In addition to differences in social, lifestyle and structural determinants of health, there is accumulating evidence for a biologic contribution to racial/ethnic disparity in cancer. To understand further the biologic mechanisms underlying racial/ethnic disparity in cancer, the analysis of differential aggregate gene expression and mutation among cancer patients of different population groups is important. Our recently published work reported differential RNA splicing as a critical mechanism underlying prostate cancer aggressiveness and drug response in African American (AA) patients. Here, we use R to analyze the Genomic Data Commons for differential aggregate gene expression (2-fold mean change, p < 0.001, Wilcoxon rank sum test) and TCGASpliceSeq to analyze TCGA for differential RNA splicing (20% median change, percent spliced in) between breast (BRCA, 183 AA and 761 white), colon (COAD, 56 AA and 194 white), squamous and adenocarcinoma lung (LUSC and LUAD, 34 and 57 AA and 392 and 444 white, respectively), and prostate (PRAD, 49 AA and 307 white) cancer specimens from AA and white patients. Our analysis identified 698 differentially expressed genes (DEGs) and 114 alternative RNA splicing events (ARSs) in BRCA, 107 DEGs and 57 ARSs in COAD, 25 and 117 DEGs and 115 and 53 ARSs in LUSC and LUAD, respectively, and 25 DEGs and 71 ARSs in PRAD from AA and white patients. Notably, comparing the lists of DEGs or ARSs from each cancer type with each of the other cancer types yields a minority of overlapping DEGs or ARSs among cancer types, indicating that race-related DEGs and ARSs are largely specific to cancer type. Pathway analysis of DEGs and ARSs reveals that the majority of these genes function in pathways relevant to cancer development and progression, such as programmed cell death, DNA repair, signal transduction, gene expression and metabolism. These analyses increase understanding of molecular mechanisms underlying racial/ethnic disparity in cancer. Upon further study of the function of these variants, such DEGs and ARSs have the potential to become candidates for development of new precision medicine interventions. Citation Format: Muthana Al Abo, Daniel J. George, Jennifer A. Freedman, Steven R. Patierno. Identification of differentially expressed and spliced genes between breast, colon, lung, and prostate cancer from African American and white patients [abstract]. In: Proceedings of the Eleventh AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2018 Nov 2-5; New Orleans, LA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2020;29(6 Suppl):Abstract nr B050.

This study aimed to identify transcriptome differences between distinct or transitional stage spherical, ovoid, and tubular porcine blastocysts throughout the initiation of elongation. We performed a global transcriptome analysis of differential gene expression using RNA‐Seq with high temporal resolution between spherical, ovoid, and tubular stage blastocysts at specific sequential stages of development from litters containing conceptus populations of distinct or transitional blastocysts. After RNA‐Seq analysis, significant differentially expressed genes (DEGs) and pathways were identified between distinct morphologies or sequential development stages. Overall, 1898 significant DEGs were identified between distinct spherical and ovoid morphologies, with 311 total DEGs between developmental stages throughout this first morphological transition, while 15 were identified between distinct ovoid and tubular, with eight total throughout these second morphological transition developmental stages. The high quantity of DEGs and pathways between conceptus stages throughout the spherical to ovoid transition suggests the importance of gene regulation during this first morphological transition for initiating elongation. Further, extensive DEG coverage of known elongation signaling pathways was illustrated from spherical to ovoid, and regulation of lipid signaling and membrane/ECM remodeling across these early conceptus stages were implicated as essential to this process, providing novel insights into potential mechanisms governing this rapid morphological change.

Background While the genetic cause of Huntington’s Disease (HD) is known since 1993, still no cure exists. Therapeutic development would benefit from a method to monitor disease progression and treatment efficacy, ideally using blood biomarkers. We previously showed that HD specific functional signatures in human blood adequately represent signatures in human brain and hence could be used as biomarkers. Aims Since potential drugs are first screened in rodent models, we aimed to determine whether the previously identified human signatures are also present in the YAC128 HD mouse model. Methods We isolated and sequenced RNA from blood collected at 12 and 20 months and four end stage brain regions from 8 YAC128 mice and 8 wild type mice. Differential gene expression analysis was applied to identify genes differentially expressed (DE) and weighted gene coexpression network analysis to identify groups of genes strongly co-expressed (modules). To technically validate RNAseq results, qPCR and western blot were performed. Results RNAseq data was validated by qPCR and western blot, confirming our gene expression analysis. Early stage blood displayed modest changes related to immune response(7 DE genes; 2 modules). At 20 months, an intermediate pathology was detected in blood (162 DE genes; 22 modules), including additional processes such as autophagy, protein transport and modification and DNA repair. In terms of differential gene expression, cortex and brainstem exhibited a mild phenotype (33 and 60 DE genes respectively), while cerebellum and striatum showed intermediate to moderate changes (145 and 101 DE genes respectively). Cerebellum and striatum showed respectively 26 and 11 modules significantly associated with HD, while this was 16 for cortex and 14 for brainstem. Representative annotations presented by all four brain regions were immune response, DNA repair, protein transport, chromatin modification and myelination. Conclusions Similar modules were present in blood and brain gene expression data from mouse and human related to immune response, protein transport and chromatin remodelling. Our next step is to statistically determine similarities between blood and brain signatures in mouse with a computational randomization experiment.

Gastrointestinal nematode (GIN) infections, particularly caused by Haemonchus contortus, are a major concern in sheep farming, resulting in significant economic losses Genetic resistance, used alongside chemical control, nutrition, and pasture management, provides a sustainable strategy to mitigate these infections in industrial livestock systems The South African Dohne Merino sheep, known for their resilience and suitability for both wool and meat production, offer a potential genetic resource for breeding GIN resistance, providing an alternative to chemical control. This study utilised RNA-Seq and differential gene expression (DEG) analysis to investigate the molecular mechanisms underlying H. contortus infection in Dohne Merino sheep. Six adult ewes from Wauldby farm (Eastern Cape Province), naturally exposed to H. contortus, were selected. The animals were categorised into resistant (n = 3) and susceptible (n = 3) groups based on Estimated Breeding Values (EBVs) from faecal egg count (FEC) phenotypes. DEG analysis revealed 34 significantly differentially expressed genes (DEGs) related to immune responses and external stimuli, with involvement in key pathways such as Rap1 and PI3K-Akt signaling, linked to H. contortus resistance. Additionally, segment-specific analysis of the gastrointestinal tract identified DEGs in different regions: 146 in the abomasum, 302 in the ileum, 584 in the jejunum, and 332 in the duodenum. The findings highlight genes and pathways contributing to GIN resistance and tissue-specific response mechanisms. These insights can support the selection and breeding of sheep with enhanced resistance to H. contortus, offering a genetic approach to combating GIN infections.

Transcriptional features of autoimmune diabetes: differentially expressed genes in CD4+ and CD8+ subpopulations of blood T cells

Cancer Pathway Connectivity Resolved By Drug Perturbation and RNA Sequencing

BackgroundEpigenome-wide association studies (EWAS) and differential gene expression analyses are generally performed on tissue samples, which consist of multiple cell types. Cell-type-specific effects of a trait, such as disease, on the omics expression are of interest but difficult or costly to measure experimentally. By measuring omics data for the bulk tissue, cell type composition of a sample can be inferred statistically. Subsequently, cell-type-specific effects are estimated by linear regression that includes terms representing the interaction between the cell type proportions and the trait. This approach involves two issues, scaling and multicollinearity.ResultsFirst, although cell composition is analyzed in linear scale, differential methylation/expression is analyzed suitably in the logit/log scale. To simultaneously analyze two scales, we applied nonlinear regression. Second, we show that the interaction terms are highly collinear, which is obstructive to ordinary regression. To cope with the multicollinearity, we applied ridge regularization. In simulated data, nonlinear ridge regression attained well-balanced sensitivity, specificity and precision. Marginal model attained the lowest precision and highest sensitivity and was the only algorithm to detect weak signal in real data.ConclusionNonlinear ridge regression performed cell-type-specific association test on bulk omics data with well-balanced performance. The omicwas package for R implements nonlinear ridge regression for cell-type-specific EWAS, differential gene expression and QTL analyses. The software is freely available from https://github.com/fumi-github/omicwas

559 Background: MAF amplification has been associated with increased bone metastases in breast cancer (BC). On the contrary, patients without MAF amplification in the primary tumor are more likely to benefit from adjuvant bisphosphonates as shown in a retrospective analysis of the AZURE trial and confirmed with a subset of NSABP-B34 specimens. A genomic signature could identify patients that lack MAF amplification as candidates for adjuvant bisphosphonates. Here we investigated the genes that could predict MAF amplification status. As MAF amplification is associated with high risk of bone metastases, 70-gene risk of distant recurrence signature (MammaPrint/MP) and 80-gene molecular subtyping signature (BluePrint/BP) were used to stratify the patient groups. Methods: A total of 166 BC patients from the UPMC were included in this pilot cohort. Fluorescence in situ hybridization was performed to detect MAF copy number. Signal-to-nucleus ratio (SNR) of ≥2.5 was used as the MAF-amplified (MAF+) cut-off. Differential gene expression analysis was performed with R limma using whole genome microarray data. MAF+ and MAF- (SNR<2.5) were compared within all patients and within patients matched by MP/BP to balance high risk groups. Differentially expressed genes (DEGs) were defined as absolute fold change ≥2 and adjusted p-value <0.05. Prediction of MAF amplification based on gene expression was performed using a correlation-based metric with training set, as well as with 1179 stage I-III BC patients from the FLEX Study (NCT03053193), which includes MP/BP testing and whole transcriptome data collection. Results: Of the 166 patients, 12% were MAF+ and 88% were MAF-. Among the MAF+ patients, 95% were MP High Risk, as expected from the association of MAF amplification and bone metastasis, as opposed to 29% of MAF- patients. Notably, there was no significant correlation between amplification and gene expression of MAF, which emphasizes the importance of utilizing other genes to predict MAF amplification. Comparing whole transcriptome of MAF+ and MAF- patients, 48 DEGs were found. From the MP/BP matched comparisons, genes ≥2-fold change were included in the final set of 57 genes, where C-X-C motif chemokine ligand and S100 calcium binding protein encoding genes were enriched. The 57-gene classifier of MAF status yielded 92% accuracy, 94% specificity, and 75% sensitivity on the training set. Interestingly, when the classifier was applied on the FLEX cohort, 12% MAF+ cases were identified, similar to the training set. Conclusions: Whole transcriptome analysis showed that BC tumors with MAF amplification are transcriptionally different than those without. Here we provide a set of 57 genes that could potentially predict MAF amplification status. Future work will expand the dataset and further explore the predictive value of such genomic signature in response to bisphosphonates. Clinical trial information: NCT03053193.

To integrate transcriptomic and genomic data to identify lung cancer biomarkers, evaluate their prognostic significance, and explore their potential in guiding personalized therapy. We integrated transcriptomic and genomic data to uncover genes associated with lung cancer risk using differential expression and Mendelian randomization (MR) analyses. Gene expression profiles from three GEO datasets (GSE19804, GSE18842, GSE19188) were batch-corrected with ComBat and analyzed to identify differentially expressed genes (DEGs). Two-sample MR was performed using lung cancer GWAS summary statistics. Functional enrichment, immune infiltration, survival analysis, and drug sensitivity prediction were conducted. The results were validated using The Cancer Genome Atlas (TCGA) lung cancer cohort. A total of 1,193 DEGs (409 upregulated, 784 downregulated) were detected between examined groups. MR analysis revealed 253 genes associated with lung cancer risk, including 118 increasing and 135 decreasing genes. Eleven genes overlapped between DEGs and MR results, with SAPCD2 and SPP1 being found to be associated with increased risk of lung cancer, whereas GPX3, C1orf162, CTSW, TPSAB1, CD93, HK3, TFPI, NEBL, and STEAP4 exhibiting inverse association. Enrichment analysis highlighted functions in vesicle lumen, carbohydrate metabolism, and immune processes. SAPCD2 and SPP1 were correlated with pro-tumor immune infiltration. Seven genes were associated with overall survival, and a derived prognostic score effectively stratified patients and correlated with TIDE and drug response profiles. This integrative genomic-transcriptomic analysis uncovered genes with suggestive causal and prognostic relevance in lung cancer, emphasizing their immune roles and therapeutic potential. These insights advance the understanding of lung cancer biology and support precision oncology strategies.

Background Chondrocytes, the only native cell type in cartilage, use mechanosensitive ion channels such as Transient Receptor Potential Vanilloid 4 (TRPV4) and PIEZO1 to transduce mechanical forces into transcriptomic changes that regulate cell behavior under both physiologic and pathologic conditions. Recent work has identified and characterized the differentially expressed genes (DEGs) that are upregulated following TRPV4 or PIEZO1 activation, but the transcriptomic systems downregulated by these ion channels also represent an important aspect of the chondrocyte regulatory process that remains poorly studied. Methods Here, we utilized previously established bulk RNAsequencing libraries to analyze the transcriptomes downregulated by activation of TRPV4 and PIEZO1 through differential gene expression analysis (using DESeq2), Gene Ontology, RT-qPCR, and Weighted Gene Correlation Network Analysis (WGCNA). Results TRPV4 and PIEZO1 activations downregulated largely unique sets of DEGs, though the set of DEGs downregulated by TRPV4 exhibited a notable overlap with genes downregulated by treatment with inflammatory mediator Interleukin-1 (IL-1). The DEG set downregulated by PIEZO1 activation included genes associated with the G2/M cell cycle checkpoint, a system that checks cells for DNA damage prior to entry into mitosis, and this result was confirmed with RT-qPCR. WGCNA revealed modules of gene regulation negatively correlated with TRPV4, PIEZO1, and IL-1, outlining how these downregulated DEGs may interact to form gene regulatory networks (GRNs). Conclusion This study complements previous work in describing the full mechanosensitive transcriptome (or “mechanome”) of differential gene expression in response to activation of mechanosensitive ion channels TRPV4 and PIEZO1 Q2 and suggests potential avenues for future therapeutic treatment design.

Fungal keratitis, caused primarily by Fusarium spp. and Aspergillus flavus, is a significant cause of corneal blindness, particularly in tropical regions. Current antifungal agents like natamycin and voriconazole have limited efficacy, underscoring the need for a deeper understanding of host immune responses. This study employed high-throughput RNA sequencing to investigate differential gene expression in human corneal tissues from patients with Fusarium spp. and A. flavus keratitis and compared them to control cadaver corneal samples. RNA was extracted from infected and control samples, followed by sequencing and differential expression analysis. Further confirmation of differential expression of selected genes were carried out by Real-Time quantitative PCR (RT-qPCR). Data analysis identified common and Fusarium spp. and A. flavus-specific differentially expressed genes (DEGs). Pathway enrichment analysis using common genes identified pathways enriched in both infections, such as interleukin 17 (IL-17), tumor necrosis factor (TNF), and chemokine signalling. Expression of hub genes, including S100 calcium binding protein A7 (S100A7), S100 calcium binding protein A8 (S100A8), S100 calcium binding protein A9 (S100A9) and C-X-C motif chemokine ligand 8 (CXCL8), identified in interleukin 17 (IL-17) signalling, was confirmed by RT-qPCR analysis. Fusarium spp.-specific DEGs, including complement C3 (C3), interleukin 6 (IL-6), interleukin 19 (IL-19) and leucine rich alpha-2-glycoprotein 1 (LRG1), are enriched in pathways such as positive regulation of immune responses, acute inflammatory responses, leukocyte cell-cell adhesion, and the regulation of cell-cell adhesion. A. flavus-specific DEGs, such as triggering receptor expressed on myeloid cells 2 (TREM2) and apolipoprotein E (APOE), are predominantly enriched in adaptive immune response, negative regulation of immune system process, negative regulation of immune response, cell migration and motility pathways. RT-qPCR confirmed the key pathogen-specific DEGs, highlighting their potential as biomarkers for pathogen-specific immune responses. These findings provide insights into the distinct immune pathways triggered by Fusarium spp. and A. flavus, offering new therapeutic targets for improving fungal keratitis treatment.

The confounding effects of high genetic diversity on the determination and interpretation of differential gene expression analysis in the parasitic nematode Haemonchus contortus

Study question Which genes are differentially expressed between patients with and without testicular fibrosis? Summary answer This study revealed three X-related genes MXRA5, DCX and VC3BX, which may be involved in Klinefelter-related testicular fibrosis. What is known already Klinefelter syndrome (KS; 47,XXY) affects 1-2 in 1000 males. Most KS men suffer from azoospermia due to a loss of spermatogonial stem cells. Additionally, testicular fibrosis is detected from puberty onwards. However, mechanisms responsible for fibrosis and germ cell loss remain unknown. Previous genomics studies on KS tissue focused on germ cell loss, however, differential gene expression analyses focused on testicular fibrosis have not been performed before. This study aimed to identify factors involved in the fibrotic remodeling of KS testes by analyzing the transcriptome of (non-)fibrotic testicular tissue. Study design, size, duration Transcriptome analysis on extracted RNA from testicular biopsies was performed. RNA scope analysis and immunohistochemistry were performed as validation for the findings of the transcriptomics study. Participants/materials, setting, methods RNA sequencing was performed to compare the genetic profile of testicular biopsies from patients with (KS and testis atrophy) and without (Sertoli cell-only syndrome and fertile controls) testicular fibrosis (n = 5, each). Next, differentially expressed genes (DEGs) between KS and testis atrophy samples were compared. To gain insight in potential functions of DEGs (significant when p < 0.01 and log2FC > 2), gene-ontology and KEGG analyses were performed. To validate the gene expression results, immunohistochemistry and RNA scope were performed. Main results and the role of chance A first transcriptomic analysis of fibrotic versus non-fibrotic testis tissue resulted in 734 significant DEGs (167 up- and 567 downregulated), of which 26 were X-linked. In the top upregulated biological functions, DEGs involved in the extracellular structure organization were found, including vascular cell adhesion molecule 1 (VCAM1). KEGG analysis showed an upregulation of genes involved in the TGF-β pathway. The second analysis of KS versus testis atrophy samples resulted in 539 significant DEGs (59 up- and 480 downregulated). One of the biological functions found though gene ontology analysis was the chronic inflammatory response. When looking at the overlap of DEGs on the X-chromosome from the first and second analysis, three genes were found: matrix-remodeling associated 5 (MXRA5), doublecortin (DCX) and variable charge X-Linked 3B (VCX3B). Through validation by immunohistochemistry and RNA scope, an overexpression of VCAM1, MXRA5 and DCX was found within the fibrotic group compared to the non-fibrotic group. Limitations, reasons for caution The study included fresh testis tissue from adult KS patients, however these are quite scarce, resulting in a low number of included patients per group (n = 5). Wider implications of the findings This study revealed genes which may play a role in testicular fibrosis, including VCAM1. In addition, fibrotic genes on the X-chromosome were revealed: MXRA5, DCX and VCX3B. Up- or downregulation of these genes may prevent testicular fibrosis and thus enhance the chances at retrieving spermatozoa from KS patients. Trial registration number NA