Identification Of Gene Signature Research Articles

Gene signatures for cancer research: A 25-year retrospective and future avenues.

Over the past two decades, extensive studies, particularly in cancer analysis through large datasets like The Cancer Genome Atlas (TCGA), have aimed at improving patient therapies and precision medicine. However, limited overlap and inconsistencies among gene signatures across different cohorts pose challenges. The dynamic nature of the transcriptome, encompassing diverse RNA species and functional complexities at gene and isoform levels, introduces intricacies, and current gene signatures face reproducibility issues due to the unique transcriptomic landscape of each patient. In this context, discrepancies arising from diverse sequencing technologies, data analysis algorithms, and software tools further hinder consistency. While careful experimental design, analytical strategies, and standardized protocols could enhance reproducibility, future prospects lie in multiomics data integration, machine learning techniques, open science practices, and collaborative efforts. Standardized metrics, quality control measures, and advancements in single-cell RNA-seq will contribute to unbiased gene signature identification. In this perspective article, we outline some thoughts and insights addressing challenges, standardized practices, and advanced methodologies enhancing the reliability of gene signatures in disease transcriptomic research.

Identification of Gene Signatures Associated with COVID-19 across Children, Adolescents, and Adults in the Nasopharynx and Peripheral Blood by Using a Machine Learning Approach.

Significant variations in immune profiles across different age groups manifest distinct clinical symptoms and prognoses in Coronavirus Disease 2019 (COVID-19) patients. Predominantly, severe COVID-19 cases that require hospitalization occur in the elderly, with the risk of severe illness escalating with age among young adults, children, and adolescents. This study aimed to delineate the unique immune characteristics of COVID-19 across various age groups and evaluate the feasibility of detecting COVID-19-induced immune alterations through peripheral blood analysis. By employing a machine learning approach, we analyzed gene expression data from nasopharyngeal and peripheral blood samples of COVID-19 patients across different age brackets. Nasopharyngeal data reflected the immune response to COVID-19 in the upper respiratory tract, while peripheral blood samples provided insights into the overall immune system status. Both datasets encompassed COVID-19 patients and healthy controls, with patients divided into children, adolescents, and adult age groups. The analysis included the expression levels of 62,703 genes per patient. Then, 9 feature-sequencing methods (least absolute shrinkage and selection operator, light gradient boosting machine, Monte Carlo feature selection, random forest, ridge regression, adaptive boosting, categorical boosting, extremely randomized trees, and extreme gradient boosting) were employed to evaluate the association of the genes with COVID-19. Key genes were then utilized to develop efficient classification models. The findings identified specific markers: insulin-like growth factor binding protein 3 (downregulated in the peripheral blood of COVID-19 patients), interferon alpha-inducible protein 27 (upregulated), and SERPING1 (upregulated in nasopharyngeal tissues). In addition, fibulin-2 was downregulated in adolescent patients, but upregulated in the other groups, while epoxide hydrolase 3 was upregulated in healthy controls, but downregulated in children and adolescents. This study offers valuable insights into the local and systemic immune responses of COVID-19 patients across age groups, aiding in identifying potential therapeutic targets and formulating personalized treatment strategies.

A common gene signature of the right ventricle in failing rat and human hearts

The molecular mechanisms of progressive right heart failure are incompletely understood. In this study, we systematically examined transcriptomic changes occurring over months in isolated cardiomyocytes or whole heart tissues from failing right and left ventricles in rat models of pulmonary artery banding (PAB) or aortic banding (AOB). Detailed bioinformatics analyses resulted in the identification of gene signature, protein and transcription factor networks specific to ventricles and compensated or decompensated disease states. Proteomic and RNA-FISH analyses confirmed PAB-mediated regulation of key genes and revealed spatially heterogeneous mRNA expression in the heart. Intersection of rat PAB-specific gene sets with transcriptome datasets from human patients with chronic thromboembolic pulmonary hypertension (CTEPH) led to the identification of more than 50 genes whose expression levels correlated with the severity of right heart disease, including multiple matrix-regulating and secreted factors. These data define a conserved, differentially regulated genetic network associated with right heart failure in rats and humans.

Abstract CT085: Targeting androgen signaling in urothelial carcinoma

Abstract Background: Men are 3-4 times more likely to develop urothelial carcinoma (UC) than women, even after accounting for differences in smoking and occupational exposures. A potential cause for this difference may be androgen signaling. Testosterone receptors are present in up to 53% of UC in men and women. Androgen signaling promotes proliferation and chemotherapy resistance, while blocking androgen has been shown to sensitize UC to chemotherapy. When exposed to a chemical carcinogen, castrated male mice and androgen receptor (AR) knockout mice are less likely to develop UC. Additionally, men taking testosterone lowering agents for prostate cancer, who are concurrently treated for UC, have a 50% lower UC recurrence rate than men with normal testosterone signaling. In a study of 9 men and 1 woman with metastatic UC, all patients were treated with gemcitabine/cisplatin and concurrent enzalutamide. Patients’ testosterone levels were not medically lowered. Importantly, dihydrotestosterone, the native ligand for the AR, has higher affinity for the AR than enzalutamide. Interestingly, the only woman in the study (with presumably low dihydrotestosterone levels) had a complete response to therapy. No other complete responses were observed. The standard therapy for non-metastatic muscle invasive UC is chemotherapy followed by cystectomy. Currently this results in about a 35% pathologic complete response (pCR) rate. pCR is associated with improved overall survival. In our study (TASUC; NCT05839119), we add degarelix to standard neoadjuvant chemotherapy to treat men and women who have AR positive UC. Methods: The main eligibility criteria are: (1) Muscle-invasive urothelial carcinoma of the bladder (2) AR positive disease by nuclear staining on immunohistochemistry (3) pT2 - T4, N0 - N1, M0 (Stage II or IIIA disease) (4) Eligible for cisplatin-based chemotherapy (5) Eligible for cystectomy In this single-arm study, all patients receive standard of care gemcitabine/cisplatin neoadjuvant chemotherapy (4 cycles) with the addition of monthly degarelix during chemotherapy. The primary outcome is pCR rate at cystectomy. We are employing a Simon 2-stage minmax design. In stage I we will enroll 20 patients. If there are at least 7 pCRs (35%) we will enroll an additional 12 patients in stage II, for a total of 32 patients. This design will provide 80% statistical power to detect a change in pCR rate from 35% to 50% with a significance level of less than 0.20. If we observe at least 14 pCRs (of 32 patients) the study will be considered positive and we will plan a follow-up randomized, controlled phase II (or phase III) study. Exploratory outcomes include correlation of AR positivity with proliferation at diagnosis, identification of oncogenic androgen response elements from matched pre- and post- treatment tissue, and identification of gene signatures associated with pCR. We will also measure 2- and 5-year disease free survival. TASUC opened to accrual on October 2, 2023 and is actively recruiting. Citation Format: Mojisola Araoye, Anthony Mega, Galina G. Lagos, Benedito Carneiro, Sari Khaleel, Dragan Golijanin, SHELDON L. HOLDER. Targeting androgen signaling in urothelial carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(7_Suppl):Abstract nr CT085.

Tissue-specific reprogramming leads to angiogenic neutrophil specialization and tumor vascularization in colorectal cancer.

Neutrophil (PMN) tissue accumulation is an established feature of ulcerative colitis (UC) lesions and colorectal cancer (CRC). To assess the PMN phenotypic and functional diversification during the transition from inflammatory ulceration to CRC we analyzed the transcriptomic landscape of blood and tissue PMNs. Transcriptional programs effectively separated PMNs based on their proximity to peripheral blood, inflamed colon, and tumors. In silico pathway overrepresentation analysis, protein-network mapping, gene signature identification, and gene-ontology scoring revealed unique enrichment of angiogenic and vasculature development pathways in tumor-associated neutrophils (TANs). Functional studies utilizing ex vivo cultures, colitis-induced murine CRC, and patient-derived xenograft models demonstrated a critical role for TANs in promoting tumor vascularization. Spp1 (OPN) and Mmp14 (MT1-MMP) were identified by unbiased -omics and mechanistic studies to be highly induced in TANs, acting to critically regulate endothelial cell chemotaxis and branching. TCGA data set and clinical specimens confirmed enrichment of SPP1 and MMP14 in high-grade CRC but not in patients with UC. Pharmacological inhibition of TAN trafficking or MMP14 activity effectively reduced tumor vascular density, leading to CRC regression. Our findings demonstrate a niche-directed PMN functional specialization and identify TAN contributions to tumor vascularization, delineating what we believe to be a new therapeutic framework for CRC treatment focused on TAN angiogenic properties.

Abstract 379: Landscape of HIF-1α expression across 24,186 solid tumors using comprehensive immune profiling

Abstract Background: Transcription factor Hypoxia Inducible Factor-1α (HIF-1α) is a key regulator of various cellular and systemic responses to hypoxia, with HIF-1α mediated pathways playing an important role in tumor progression and metastasis. Recent studies have highlighted the correlation between HIF-1α expression with tumor cell survival, angiogenesis, and invasion. In this study, we explore the landscape of HIF-1α expression across 35 solid tumor histologies, its co-expression with immune checkpoint and angiogenesis-related genes, as well as its association with outcomes from a retrospective cohort of patients with non-small cell lung cancer (NSCLC) treated with pembrolizumab. Methods: A discovery cohort of 24,186 solid tumors across 35 types were evaluated by comprehensive immune profiling assay. HIF-1α gene expression was normalized and ranked against a reference population of 735 samples yielding a percentile rank values [0-100], with rank classified as high (≥75), moderate (≥25 and <75) and low (<25). We investigated the distribution of HIF-1α expression, cellular proliferation signature, and co-expression with other immunotherapy targets using Spearman correlations (rs). Additionally, Kaplan-Meier analysis and Chi-squared tests were utilized for overall survival (OS), progression free survival (PFS) and objective response rate (ORR) differences in a retrospective cohort of 72 patients with NSCLC, treated with pembrolizumab. Results: Pan-cancer analysis confirmed a broad distribution of HIF-1α expression, with highest median expression levels observed in brain and central nervous system (CNS) (median=64), thyroid (median=49) and uterine (median=42) cancers. Significant associations (p<0.001) were observed between PD-L1 positivity (TPS≥1%, CPS≥1) and HIF-1α expression, with 60% of HIF-1α high cases (n=1,314/2,193) being PD-L1 positive, whereas 53% of HIF-1α low cases (n=4,834/9,135) were PD-L1 negative. Significant correlations (p<0.05) of ranked expression of HIF-1α with immune checkpoint blockage targets like PD-1 and CTLA-4, as well as angiogenesis-related genes like VEGFA and PTEN were observed. Within our pembrolizumab-treated retrospective cohort, there were no significant differences in OS and PFS between HIF-1α high and HIF-1α low tumors. However, the response to pembrolizumab was significantly higher in HIF-1α low tumors (ORR=58.82%, p<0.01) compared to HIF-1α high tumors. Conclusion: In clinically tested solid tumors, HIF-1α showed a dynamic expression range, with highest levels observed in brain, thyroid and uterine cancers. Significant co-expression of HIF-1α with various downstream genes, especially angiogenesis-related genes, enabled identification of gene signatures that characterize the hypoxic tumor microenvironment and may aid in the development of novel combination therapy strategies. Citation Format: Hardik I. Parikh, Sarabjot Pabla, Maria-Fernanda Senosain, Robert Seager, Erik Vanroey, Shuang Gao, Yamuna Pulivendula, Paul DePietro, Mary Nesline, Durgapras Dash, Jeffrey Conroy, Stephanie Hastings, Heidi Ko, Kyle Strickland, Rebecca Previs, Eric Severson, Taylor Jensen, Shakti Ramkissoon, Marcia Eisenberg, Brian Caveney. Landscape of HIF-1α expression across 24,186 solid tumors using comprehensive immune profiling [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 379.

Abstract 7365: Drug-centric prior improves drug response signature identification in partially overlapping, large-scale pharmacogenomic datasets

Abstract Introduction: With the accumulation of large-scale pharmacogenomic data such as whole-genome RNA sequencing, copy number and mutation profiles for tens of thousands of samples, further screened with thousands of small molecules and other perturbagens, the question arises how to best leverage partially overlapping datasets generated at different facilities. Notably, Haibe-Kains et al. observed discordance in state-of-the-art pharmacogenomic repositories. As research groups across the world continue to generate drug screens of variable size and quality, the need for approaches that can learn from such partially overlapping experiments and improve the signal to noise ratio emerges with increasing importance. Methods: A previously published Bayesian group factor analysis model was shown to outperform other approaches in predicting drug response and identifying gene signature from pharmacogenomic datasets particularly by leveraging shared information for a given gene across multiple omics assays. Here, we applied the same model in a similar fashion to now learn from shared observations for the same drug across multiple partially overlapping and noisy small molecule screens. We integrate gene expression, mutation and drug response data from the two largest pan-cancer repositories at the Broad and Sanger Institutes, respectively. We train joint models on partially overlapping data from both. We evaluate our performance in three ways: 1) we test if the joint model improves the prioritization of known consensus biomarkers for the drugs shared between the two cohorts; 2) we test if the joint model improves the recapitulation of shared drug mechanism of action as compared to the single dataset models; 3) we evaluate the joint models with respect to pathway enrichment as compared to the single dataset models. Results: We evaluated the performance of our joint model for 5 drugs shared between the 2 resources: selumetinib, tanespimycin, nutlin 3A, mirdametinib and PLX4720. First, we show that training joint models on partially overlapping pharmacogenomic datasets can overall improve gene signature identification by improving the ranks of known consensus biomarkers. Second, we show that the joint model learns a latent representation of the drugs that better recapitulates the underlying known mechanisms of action for the three serine/threonine kinase inhibitors. Finally, we show that the joint model achieved improved pathway enrichment results for the targeted MAPK/ERK signaling pathway. Conclusions: We present an application of a Bayesian group factor analysis model, where we employ a drug-centric prior to transfer information about drugs screened in multiple datasets. We show that joint models leveraging partially overlapping large-scale pharmacogenomic datasets from the Broad and Sanger institutes can overall improve drug signature identification. Citation Format: Dharani Thirumalaisamy, Sunil K. Joshi, Mehmet Gönen, Olga Nikolova. Drug-centric prior improves drug response signature identification in partially overlapping, large-scale pharmacogenomic datasets [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7365.

Identification of gene signatures and molecular mechanisms underlying the mutual exclusion between psoriasis and leprosy

Leprosy and psoriasis rarely coexist, the specific molecular mechanisms underlying their mutual exclusion have not been extensively investigated. This study aimed to reveal the underlying mechanism responsible for the mutual exclusion between psoriasis and leprosy. We obtained leprosy and psoriasis data from ArrayExpress and GEO database. Differential expression analysis was conducted separately on the leprosy and psoriasis using DEseq2. Differentially expressed genes (DEGs) with opposite expression patterns in psoriasis and leprosy were identified, which could potentially involve in their mutual exclusion. Enrichment analysis was performed on these candidate mutually exclusive genes, and a protein–protein interaction (PPI) network was constructed to identify hub genes. The expression of these hub genes was further validated in an external dataset to obtain the critical mutually exclusive genes. Additionally, immune cell infiltration in psoriasis and leprosy was analyzed using single-sample gene set enrichment analysis (ssGSEA), and the correlation between critical mutually exclusive genes and immune cells was also examined. Finally, the expression pattern of critical mutually exclusive genes was evaluated in a single-cell transcriptome dataset. We identified 1098 DEGs in the leprosy dataset and 3839 DEGs in the psoriasis dataset. 48 candidate mutually exclusive genes were identified by taking the intersection. Enrichment analysis revealed that these genes were involved in cholesterol metabolism pathways. Through PPI network analysis, we identified APOE, CYP27A1, FADS1, and SOAT1 as hub genes. APOE, CYP27A1, and SOAT1 were subsequently validated as critical mutually exclusive genes on both internal and external datasets. Analysis of immune cell infiltration indicated higher abundance of 16 immune cell types in psoriasis and leprosy compared to normal controls. The abundance of 6 immune cell types in psoriasis and leprosy positively correlated with the expression levels of APOE and CYP27A1. Single-cell data analysis demonstrated that critical mutually exclusive genes were predominantly expressed in Schwann cells and fibroblasts. This study identified APOE, CYP27A1, and SOAT1 as critical mutually exclusive genes. Cholesterol metabolism pathway illustrated the possible mechanism of the inverse association of psoriasis and leprosy. The findings of this study provide a basis for identifying mechanisms and therapeutic targets for psoriasis.

Associating growth factor secretions and transcriptomes of single cells in nanovials using SEC-seq.

Cells secrete numerous bioactive molecules that are essential for the function of healthy organisms. However, scalable methods are needed to link individual cell secretions to their transcriptional state over time. Here, by developing and using secretion-encoded single-cell sequencing (SEC-seq), which exploits hydrogel particles with subnanolitre cavities (nanovials) to capture individual cells and their secretions, we simultaneously measured the secretion of vascular endothelial growth factor A (VEGF-A) and the transcriptome for thousands of individual mesenchymal stromal cells. Our data indicate that VEGF-A secretion is heterogeneous across the cell population and is poorly correlated with the VEGFA transcript level. The highest VEGF-A secretion occurs in a subpopulation of mesenchymal stromal cells characterized by a unique gene expression signature comprising a surface marker, interleukin-13 receptor subunit alpha 2 (IL13RA2), which allowed the enrichment of this subpopulation. SEC-seq enables the identification of gene signatures linked to specific secretory states, facilitating mechanistic studies, the isolation of secretory subpopulations and the development of means to modulate cellular secretion.

Optimal ranking and directional signature classification using the integral strategy of multi-objective optimization-based association rule mining of multi-omics data.

Introduction: Association rule mining (ARM) is a powerful tool for exploring the informative relationships among multiple items (genes) in any dataset. The main problem of ARM is that it generates many rules containing different rule-informative values, which becomes a challenge for the user to choose the effective rules. In addition, few works have been performed on the integration of multiple biological datasets and variable cutoff values in ARM. Methods: To solve all these problems, in this article, we developed a novel framework MOOVARM (multi-objective optimized variable cutoff-based association rule mining) for multi-omics profiles. Results: In this regard, we identified the positive ideal solution (PIS), which maximized the profit and minimized the loss, and negative ideal solution (NIS), which minimized the profit and maximized the loss for all gene sets (item sets), belonging to each extracted rule. Thereafter, we computed the distance (d +) from PIS and distance (d -) from NIS for each gene set or product. These two distances played an important role in determining the optimized associations among various pairs of genes in the multi-omics dataset. We then globally estimated the relative closeness to PIS for ranking the gene sets. When the relative closeness score of the rule is greater than or equal to the pre-defined threshold value, the rule can be considered a final resultant rule. Moreover, MOOVARM evaluated the relative score of the rule based on the status of all genes instead of individual genes. Conclusions: MOOVARM produced the final rank of the extracted (multi-objective optimized) rules of correlated genes which had better disease classification than the state-of-the-art algorithms on gene signature identification.

AAFL: automatic association feature learning for gene signature identification of cancer subtypes in single-cell RNA-seq data.

Single-cell RNA-sequencing (scRNA-seq) technologies have enabled the study of human cancers in individual cells, which explores the cellular heterogeneity and the genotypic status of tumors. Gene signature identification plays an important role in the precise classification of cancer subtypes. However, most existing gene selection methods only select the same informative genes for each subtype. In this study, we propose a novel gene selection method, automatic association feature learning (AAFL), which automatically identifies different gene signatures for different cell subpopulations (cancer subtypes) at the same time. The proposed AAFL method combines the residual network with the low-rank network, which selects genes that are most associated with the corresponding cell subpopulations. Moreover, the differential expression genes are acquired before gene selection to filter the redundant genes. We apply the proposed feature learning method to the real cancer scRNA-seq data sets (melanoma) to identify cancer subtypes and detect gene signatures of identified cancer subtypes. The experimental results demonstrate that the proposed method can automatically identify different gene signatures for identified cancer subtypes. Gene ontology enrichment analysis shows that the identified gene signatures of different subtypes reveal the key biological processes and pathways. These gene signatures are expected to bring important implications for understanding cellular heterogeneity and the complex ecosystem of tumors.

Identification of gene signatures associated with ulcerative colitis and the association with immune infiltrates in colon cancer.

Inflammatory bowel diseases, including ulcerative colitis (UC) and Crohn's disease, are some of the most common inflammatory disorders of the gastrointestinal tract. The dysfunction of the immune system in the intestines is suggested to be the underlying cause of the pathogenesis of UC. However, the mechanisms regulating these dysfunctional immune cells and inflammatory phenotypes are still unclear. The differential expression analysis on microarray datasets were performed including GSE24287, GSE87466, GSE102133, and GSE107499, including 376 samples. "Gene Ontology" and "Kyoto Encyclopedia of Genes and Genomes" pathway enrichment analyses were conducted to identify the common differentially expressed genes (DEGs) in these datasets and explore their underlying biological mechanisms. Further algorithms like "Cell-type Identification by Estimating Relative Subsets of RNA Transcripts" were used to determine the infiltration status of immune cells in patients with UC. "Cytoscape" and "Gene Set Enrichment Analysis" were used to screen for hub genes and to investigate their biological mechanisms. The Tumor Immune Estimation Resource database was used to study the correlation between hub genes and infiltrating immune cells in patients with UC. A total of three hub genes, CCL3, MMP3, and TIMP1, were identified using Cytoscape. A positive correlation was observed between these hub genes and patients with active UC. These genes served as a biomarker for active UC. Moreover, a decrease in CCL3, MMP3, and TIMP1 expression was observed in the mucosa of the intestine of patients with active UC who responded to Golimumab therapy. In addition, results show a significant positive correlation between CCL3, MMP3, and TIMP1 expression and different immune cell types including dendritic cells, macrophages, CD8+ T cells, and neutrophils in patients with colon cancer. Moreover, CCL3, MMP3, and TIMP1 expression were strongly correlated with different immune cell markers. Study results show the involvement of hub genes like CCL3, MMP3, and TIMP1 in the pathogenesis of UC. These genes could serve as a novel pharmacological regulator of UC. These could be used as a therapeutic target for treating patients with UC and may serve as biomarkers for immune cell infiltration in colon cancer.

Identification of a gene expression signature associated with breast cancer survival and risk that improves clinical genomic platforms.

Modern genomic technologies allow us to perform genome-wide analysis to find gene markers associated with the risk and survival in cancer patients. Accurate risk prediction and patient stratification based on robust gene signatures is a key path forward in personalized treatment and precision medicine. Several authors have proposed the identification of gene signatures to assign risk in patients with breast cancer (BRCA), and some of these signatures have been implemented within commercial platforms in the clinic, such as Oncotype and Prosigna. However, these platforms are black boxes in which the influence of selected genes as survival markers is unclear and where the risk scores provided cannot be clearly related to the standard clinicopathological tumor markers obtained by immunohistochemistry (IHC), which guide clinical and therapeutic decisions in breast cancer. Here, we present a framework to discover a robust list of gene expression markers associated with survival that can be biologically interpreted in terms of the three main biomolecular factors (IHC clinical markers: ER, PR and HER2) that define clinical outcome in BRCA. To test and ensure the reproducibility of the results, we compiled and analyzed two independent datasets with a large number of tumor samples (1024 and 879) that include full genome-wide expression profiles and survival data. Using these two cohorts, we obtained a robust subset of gene survival markers that correlate well with the major IHC clinical markers used in breast cancer. The geneset of survival markers that we identify (which includes 34 genes) significantly improves the risk prediction provided by the genesets included in the commercial platforms: Oncotype (16 genes) and Prosigna (50 genes, i.e. PAM50). Furthermore, some of the genes identified have recently been proposed in the literature as new prognostic markers and may deserve more attention in current clinical trials to improve breast cancer risk prediction. All data integrated and analyzed in this research will be available on GitHub (https://github.com/jdelasrivas-lab/breastcancersurvsign), including the R scripts and protocols used for the analyses. Supplementary data are available at Bioinformatics Advances online.

A novel cuproptosis-related immune checkpoint gene signature identification and experimental validation in hepatocellular carcinoma

Copper-induced death, also termed cuproptosis, is a novel form of programmed cell death and is promising as a new strategy for cancer therapeutics. Elevated copper levels in tumor cells are positively associated with high PD-L1 expression. Nonetheless, the prognostic significance of cuproptosis-related immune checkpoint genes (CRICGs) in hepatocellular carcinoma remains to be further clarified. This study aimed to construct the prognostic CRICG signature to predict the immunotherapy response and outcomes of HCC patients. The co-expressed CRICGs were first screened through Pearson correlation analysis. Based on the least absolute shrinkage and selection operator-COX regression analyses, we identified a prognostic 5-CRICGs model, which closely correlates with poor outcomes, cancer development, and immune response to hepatocellular carcinoma. External validation was conducted using the GSE14520 dataset. Lastly, qRT-PCR was performed to determine the expression of the CRICGs in HCC. In summary, we developed and validated a novel prognostic CRICG model based on 5 CRICGs. This prognostic signature could effectively forecast the outcomes and immune response of HCC patients, which may serve as biomarkers for anticancer therapy.

Deep transfer learning of cancer drug responses by integrating bulk and single-cell RNA-seq data

Drug screening data from massive bulk gene expression databases can be analyzed to determine the optimal clinical application of cancer drugs. The growing amount of single-cell RNA sequencing (scRNA-seq) data also provides insights into improving therapeutic effectiveness by helping to study the heterogeneity of drug responses for cancer cell subpopulations. Developing computational approaches to predict and interpret cancer drug response in single-cell data collected from clinical samples can be very useful. We propose scDEAL, a deep transfer learning framework for cancer drug response prediction at the single-cell level by integrating large-scale bulk cell-line data. The highlight in scDEAL involves harmonizing drug-related bulk RNA-seq data with scRNA-seq data and transferring the model trained on bulk RNA-seq data to predict drug responses in scRNA-seq. Another feature of scDEAL is the integrated gradient feature interpretation to infer the signature genes of drug resistance mechanisms. We benchmark scDEAL on six scRNA-seq datasets and demonstrate its model interpretability via three case studies focusing on drug response label prediction, gene signature identification, and pseudotime analysis. We believe that scDEAL could help study cell reprogramming, drug selection, and repurposing for improving therapeutic efficacy.

Identification of gene signatures for COAD using feature selection and Bayesian network approaches

The combination of TCGA and GTEx databases will provide more comprehensive information for characterizing the human genome in health and disease, especially for underlying the cancer genetic alterations. Here we analyzed the gene expression profile of COAD in both tumor samples from TCGA and normal colon tissues from GTEx. Using the SNR-PPFS feature selection algorithms, we discovered a 38 gene signatures that performed well in distinguishing COAD tumors from normal samples. Bayesian network of the 38 genes revealed that DEGs with similar expression patterns or functions interacted more closely. We identified 14 up-DEGs that were significantly correlated with tumor stages. Cox regression analysis demonstrated that tumor stage, STMN4 and FAM135B dysregulation were independent prognostic factors for COAD survival outcomes. Overall, this study indicates that using feature selection approaches to select key gene signatures from high-dimensional datasets can be an effective way for studying cancer genomic characteristics.

Sequential reinforcement active feature learning for gene signature identification in renal cell carcinoma.

Renal cell carcinoma (RCC) is one of the deadliest cancers and mainly consists of three subtypes: kidney clear cell carcinoma (KIRC), kidney papillary cell carcinoma (KIRP), and kidney chromophobe (KICH). Gene signature identification plays an important role in the precise classification of RCC subtypes and personalized treatment. However, most of the existing gene selection methods focus on statically selecting the same informative genes for each subtype, and fail to consider the heterogeneity of patients which causes pattern differences in each subtype. In this work, to explore different informative gene subsets for each subtype, we propose a novel gene selection method, named sequential reinforcement active feature learning (SRAFL), which dynamically acquire the different genes in each sample to identify the different gene signatures for each subtype. The proposed SRAFL method combines the cancer subtype classifier with the reinforcement learning (RL) agent, which sequentially select the active genes in each sample from three mixed RCC subtypes in a cost-sensitive manner. Moreover, the module-based gene filtering is run before gene selection to filter the redundant genes. We mainly evaluate the proposed SRAFL method based on mRNA and long non-coding RNA (lncRNA) expression profiles of RCC datasets from The Cancer Genome Atlas (TCGA). The experimental results demonstrate that the proposed method can automatically identify different gene signatures for different subtypes to accurately classify RCC subtypes. More importantly, we here for the first time show the proposed SRAFL method can consider the heterogeneity of samples to select different gene signatures for different RCC subtypes, which shows more potential for the precision-based RCC care in the future.

Survival Genie, a web platform for survival analysis across pediatric and adult cancers

The genomics data-driven identification of gene signatures and pathways has been routinely explored for predicting cancer survival and making decisions related to targeted treatments. A large number of packages and tools have been developed to correlate gene expression/mutations to the clinical outcome but lack the ability to perform such analysis based on pathways, gene sets, and gene ratios. Furthermore, in this single-cell omics era, the cluster markers from cancer single-cell transcriptomics studies remain an underutilized prognostic option. Additionally, no bioinformatics online tool evaluates the associations between the enrichment of canonical cell types and survival across cancers. Here we have developed Survival Genie, a web tool to perform survival analysis on single-cell RNA-seq (scRNA-seq) data and a variety of other molecular inputs such as gene sets, genes ratio, tumor-infiltrating immune cells proportion, gene expression profile scores, and tumor mutation burden. For a comprehensive analysis, Survival Genie contains 53 datasets of 27 distinct malignancies from 11 different cancer programs related to adult and pediatric cancers. Users can upload scRNA-seq data or gene sets and select a gene expression partitioning method (i.e., mean, median, quartile, cutp) to determine the effect of expression levels on survival outcomes. The tool provides comprehensive results including box plots of low and high-risk groups, Kaplan–Meier plots with univariate Cox proportional hazards model, and correlation of immune cell enrichment and molecular profile. The analytical options and comprehensive collection of cancer datasets make Survival Genie a unique resource to correlate gene sets, pathways, cellular enrichment, and single-cell signatures to clinical outcomes to assist in developing next-generation prognostic and therapeutic biomarkers. Survival Genie is open-source and available online at https://bbisr.shinyapps.winship.emory.edu/SurvivalGenie/.

Acquisition of taxane resistance by p53 inactivation in ovarian cancer cells.

Ovarian cancer is one of the most common gynecologic malignancies in women and has a poor prognosis. Taxanes are a class of standard first-line chemotherapeutic agents for the treatment of ovarian cancer. However, tumor-intrinsic and acquired resistance to taxanes poses major challenges to improving clinical outcomes. Hence, there is an urgent clinical need to understand the mechanisms of resistance in order to discover potential biomarkers and therapeutic strategies to increase taxane sensitivity in ovarian cancer. Here, we report the identification of an association between the TP53 status and taxane sensitivity in ovarian cancer cells through complementary experimental and informatics approaches. We found that TP53 inactivation is associated with taxane resistance in ovarian cancer cells, supported by the evidence from (i) drug sensitivity profiling with bioinformatic analysis of large-scale cancer therapeutic response and genomic datasets and (ii) gene signature identification based on experimental isogenic cell line models. Further, our studies revealed TP53-dependent gene expression patterns, such as overexpression of ACSM3, as potential predictive biomarkers of taxane resistance in ovarian cancer. The TP53-dependent hyperactivation of the WNT/β-catenin pathway discovered herein revealed a potential vulnerability to exploit in developing combination therapeutic strategies. Identification of this genotype-phenotype relationship between the TP53 status and taxane sensitivity sheds light on TP53-directed patient stratification and therapeutic discoveries for ovarian cancer treatment.

Identification of gene signature in RNA-Seq hepatocellular carcinoma data by Pareto-optimal cluster algorithm.

This study aimed to detect gene signatures in RNA-sequencing (RNA-seq) data using Pareto-optimal cluster size identification. RNA-seq has emerged as an important technology for transcriptome profiling in recent years. Gene expression signatures involving tens of genes have been proven to be predictive of disease type and patient response to treatment. Data related to the liver cancer RNA-seq dataset, which included 35 paired hepatocellular carcinoma (HCC) and non-tumor tissue samples, was used in this study. The differentially expressed genes (DEGs) were identified after performing pre-filtering and normalization. After that, a multi-objective optimization technique, namely multi-objective optimization for collecting cluster alternatives (MOCCA), was used to discover the Pareto-optimal cluster size for these DEGs. Then, the k-means clustering method was performed on the RNA-seq data. The best cluster, as a signature for the disease, was found by calculating the average Spearman's correlation score of all genes in the module in a pair-wise manner. All analyses were performed in the R 4.1.1 package in virtual space with 100 Gb of RAM memory. Using MOCCA, eight Pareto-optimal clusters were obtained. Ultimately, two clusters with the greatest average Spearman's correlation coefficient scores were chosen as gene signatures. Eleven prognostic genes involved in HCC's abnormal metabolism were identified. In addition, three differentially expressed pathways were identified between tumor and non-tumor tissues. These identified metabolic prognostic genes help us to provide more powerful prognostic information and enhance survival prediction for HCC patients. In addition, Pareto-optimal cluster size identification is suggested for gene signature in other RNA-Seq data.