Drug Sensitivity Prediction Research Articles

Establishment of a Necroptosis-Related Prognostic Signature to Reveal Immune Infiltration and Predict Drug Sensitivity in Hepatocellular Carcinoma.

Background: Hepatocellular carcinoma (HCC) is a common type of primary liver cancer and has a poor prognosis. In recent times, necroptosis has been reported to be involved in the progression of multiple cancers. However, the role of necroptosis in HCC prognosis remains elusive. Methods: The RNA-seq data and clinical information of HCC patients were downloaded from The Cancer Genome Atlas (TCGA) and International Cancer Genome Consortium (ICGC) databases. Differentially expressed genes (DEGs) and prognosis-related genes were explored, and the nonnegative matrix factorization (NMF) clustering algorithm was applied to divide HCC patients into different subtypes. Based on the prognosis-related DEGs, univariate Cox and LASSO Cox regression analyses were used to construct a necroptosis-related prognostic model. The relationship between the prognostic model and immune cell infiltration, tumor mutational burden (TMB), and drug response were explored. Results: In this study, 13 prognosis-related DEGs were confirmed from 18 DEGs and 24 prognostic-related genes. Based on the prognosis-related DEGs, patients in the TCGA cohort were clustered into three subtypes by the NMF algorithm, and patients in C3 had better survival. A necroptosis-related prognostic model was established according to LASSO analysis, and HCC patients in TCGA and ICGC were divided into high- and low-risk groups. Kaplan–Meier (K–M) survival analysis revealed that patients in the high-risk group had a shorter survival time compared to those in the low-risk group. Using univariate and multivariate Cox analyses, the prognostic model was identified as an independent prognostic factor and had better survival predictive ability in HCC patients compared with other clinical biomarkers. Furthermore, the results revealed that the high-risk patients had higher stromal, immune, and ESTIMATE scores; higher TP53 mutation rate; higher TMB; and lower tumor purities compared to those in the low-risk group. In addition, there were significant differences in predicting the drug response between the high- and low-risk groups. The protein and mRNA levels of these prognostic genes were upregulated in HCC tissues compared to normal liver tissues. Conclusion: We established a necroptosis-related prognostic signature that may provide guidance for individualized drug therapy in HCC patients; however, further experimentation is needed to validate our results.

Establishment of a circular RNA regulatory stemness-related gene pair signature for predicting prognosis and therapeutic response in colorectal cancer.

BackgroundColorectal cancer (CRC) is a common malignant tumor of the digestive tract with a poor prognosis. Cancer stem cells (CSCs) affect disease outcomes and treatment responses in CRC. We developed a circular RNA (circRNA) regulatory stemness-related gene pair (CRSRGP) signature to predict CRC patient prognosis and treatment effects.MethodsThe circRNA, miRNA, and mRNA expression profiles and clinical information of CRC patients were obtained from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases. CRSRGPs were established based on stemness-related genes in the competing endogenous RNA (ceRNA) network. A CRSRGP signature was generated using the least absolute shrinkage and selection operator (Lasso) and Cox regression analysis of TCGA training set. The prognosis was predicted by generating a nomogram integrating the CRSRGP signature and clinicopathologic features. The model was validated in an external validation set (GSE17536). The antitumor drug sensitivity and immunotherapy responses of CRC patients in the high-risk group (HRG) and low-risk group (LRG) were evaluated by the pRRophetic algorithm and immune checkpoint analysis.ResultsWe established an 18-CRSRGP signature to predict the prognosis and treatment responses of CRC patients. In the training and external validation sets, risk scores were used to categorize CRC patients into the HRG and LRG. The Kaplan–Meier analysis showed a poor prognosis for patients in the HRG and that subgroups with different clinical characteristics had significantly different prognoses. A multivariate Cox analysis revealed that the CRSRGP signature was an independent prognostic factor. The nomogram integrating clinical features and the CRSRGP signature efficiently predicted CRC patient prognosis, outperformed the current TNM staging system, and had improved practical clinical value. Anticancer drug sensitivity predictions revealed that the tumors of patients in the HRG were more sensitive to pazopanib, sunitinib, gemcitabine, lapatinib, and cyclopamine. Analysis of immune checkpoint markers demonstrated that patients in the HRG were more likely to benefit from immunotherapy.ConclusionAn efficient, reliable tool for evaluating CRC patient prognosis and treatment response was established based on the 18-CRSRGP signature and nomogram.

EBSRMF: Ensemble based similarity-regularized matrix factorization to predict anticancer drug responses

Drug sensitivity prediction to a panel of cancer cell lines using computational approaches has been a challenge for two decades. With the emergence of high-throughput screening technologies, thousands of compounds and cancer cell lines panels with drug sensitivity data are publicly available at various pharmacogenomics databases. Analyzing these data is crucial to improve cancer treatment and develop new anticancer drugs. In this work, we propose EBSRMF: Ensemble Based Similarity-Regularized Matrix Factorization, which is a bagging based framework to improve the drug sensitivity prediction on the Cancer Cell Line Encyclopedia (CCLE) data. Based on the fact that similar drugs and cell lines exhibit similar drug response, we have investigated cell line and drug similarity matrices based on gene expression profiles and chemical structure respectively. The drug sensitivity value is used as outcome values which are the half maximal inhibitory concentrations (IC50). In order to improve the generalization ability of the proposed model, a homogeneous ensemble based bagging learning approach is also investigated where multiple SRMF models are used to train N subsets of the input data. The outcome of each training algorithm is aggregated using the averaging method to predict the outcome. Experiments are conducted on two benchmark datasets: CCLE and GDSC. The proposed model is compared with state-of-the-art models using multiple evaluation metrics including Root Means Square Error (RMSE) and Pearson Correlation Coefficient (PCC). The proposed model is quite promising and achieves better performance on CCLE dataset when compared with the existing approaches.

Hybrid Sea Lion Crow Search Algorithm-Based Stacked Autoencoder for Drug Sensitivity Prediction From Cancer Cell Lines

Providing better therapy to cancer patients remains a major task due to drug resistance of tumor cells. This paper proposes a sea lion crow search algorithm (SLCSA) for drug sensitivity prediction. The drug sensitivity from cultured cell lines is predicted using stacked autoencoder, and the proposed SLCSA is derived from a combination of sea lion optimization (SLnO) and crow search algorithm (CSA). The implemented approach has offered superior results. The maximum value of testing accuracy for normal is 0.920, leukemia is 0.920, NSCLC is 0.912, and urogenital is 0.914.

A Novel Thrombosis-Related Signature for Predicting Survival and Drug Compounds in Glioblastoma.

Glioblastoma is the most common primary tumor in the central nervous system, and thrombosis-associated genes are related to its occurrence and progression. Univariate Cox and LASSO regression analysis were utilized to develop a new prognostic signature based on thrombosis-associated genes. Gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and HALLMARK were used for functional annotation of risk signature. ESTIMATE, MCP-counter, xCell, and TIMER algorithms were used to quantify immune infiltration in the tumor microenvironment. Genomics of Drug Sensitivity in Cancer (GDSC) was used for selecting potential drug compounds. Risk signature based on thrombosis-associated genes shows moderate performance in prognosis prediction. The functional annotation of the risk signature indicates that the signaling pathways related to the cell cycle, apoptosis, tumorigenesis, and immune suppression are rich in the high-risk group. Somatic mutation analysis shows that tumor-suppressive gene TP53 and oncogene PTEN have higher expression in low-risk and high-risk groups, respectively. Potential drug compounds are explored in risk score groups and show higher AUC values in the low-risk score group. A nomogram with valuable prognostic factors exhibits high sensitivity in predicting the survival outcome of GBM patients. Our research screens out multiple thromboses-associated genes with remarkable clinical significance in GBM and further develops a meaningful prognostic risk signature predicting drug sensitivity and survival outcome.

High Expression of TACC3 Is Associated with the Poor Prognosis and Immune Infiltration in Lung Adenocarcinoma Patients.

Background Lung adenocarcinoma (LUAD) has been recognized as one of the commonest aggressive malignant tumors occurring in humans. The transforming acidic coiled-coil-containing protein 3 (TACC3) seems to be a probable prognostic marker and treatment target for non-small-cell lung cancer (NSCLC). Nevertheless, there exist no reports on the association between TACC3 and immunotherapy or other therapeutic interventions in LUAD. Methods Premised on the data accessed from The Cancer Genome Atlas- (TCGA-) LUAD, we carried out bioinformatics analysis. The TACC3 expression in LUAD was analyzed utilizing the GEPIA. A survival module was constructed to evaluate the effect of TACC3 on the survival of patients with LUAD. Logistic regression was undertaken to examine the relationship between TACC3 expression and clinical factors. Protein-protein interaction analysis was performed in the GeneMANIA database, and enrichment analysis and identification of predicted signaling pathways were performed using Gene Ontology and Kyoto Encyclopedia of Genes. Additionally, the Cox regression was used to assess the clinicopathologic features linked to the overall survival in TCGA patients. Lastly, we investigated the link between TACC3 and tumor-infiltrating immune cells (TIICs) through CIBERSORT and the “Correlation” module of GEPIA. The association between TACC3 gene expression and drug response was analyzed using the CellMiner database to predict drug sensitivity. Results The outcomes illustrated that TACC3 was upregulated and considerably correlated with dismal prognosis in LUAD patients. Moreover, the multivariate Cox regression analysis depicted TACC3 as an independent prognostic marker in LUAD patients. It was also revealed that the expression of TACC3 was related to clinical stage (P = 0.014), age (P = 0.002), and T classification (P ≤ 0.018). Moreover, we discovered that the expression of TACC3 was considerably linked to a wide range of TIICs, especially the T cells and NK cells. Single-cell results found that TACC3 was mainly expressed in the immune cells (especially tprolif cells) and malignant cells. TACC3 gene expression was positively correlated with TMB and MSI, and TACC3 may provide a prediction of the efficacy of immunotherapy. Moreover, the correlation analysis between TACC3 gene expression and immune checkpoint gene expression revealed that TACC3 may coordinate the activities of these ICP genes in different signal transduction pathways. TACC3 is related to biological progress (BP), cellular component (CC), and molecular function (MF). The pathways involved in the interaction network involving TACC3 include nonhomologous end-joining, RNA transport, pantothenate and CoA biosynthesis, homologous recombination, and nucleotide excision repair. Furthermore, we investigated the association between the expression of TACC3 and the use of antitumor drugs, and TACC3 was positively correlated with response to most drugs. Conclusion The findings from this research offer robust proof that the expression of TACC3 could be a prognostic marker correlated with TIICs in LUAD. TACC3 can also provide new ideas for immunotherapy as a potential therapeutic target.

Predicting tumor drug sensitivity with multi-omics data

The prediction of tumor drug sensitivity plays an important role in clinically guiding patients' medication. In this paper, a multi-omics data-based cancer drug sensitivity prediction model was constructed by Stacking ensemble learning method. The data including gene expression, mutation, copy number variation and drug sensitivity value (IC50) of 198 drugs were downloaded from the GDSC database. Multiple feature selection methods were applied for dimensionality reduction. Six primary learners and one secondary learner were integrated into modeling by Stacking method. The model was validated with 5-fold cross-validation. In the prediction results, 36.4% of drug models' AUCs were greater than 0.9, 49.0% of drug models' AUCs were between 0.8-0.9, and the lowest drug model's AUC was 0.682. The multi-omics model for drug sensitivity prediction based on Stacking method is better than the known single-omics or multi-omics model in terms of accuracy and stability. The model based on multi-omics data is better than the single-omics data in predicting drug sensitivity. Function annotation and enrichment analysis of feature genes revealed the potential resistance mechanism of tumors to sorafenib, providing the model interpretability from a biological perspective, and demonstrated the model's potential applicability in clinical medication guidance.

Differential Private Deep Learning Models for Analyzing Breast Cancer Omics Data.

Proper analysis of high-dimensional human genomic data is necessary to increase human knowledge about fundamental biological questions such as disease associations and drug sensitivity. However, such data contain sensitive private information about individuals and can be used to identify an individual (i.e., privacy violation) uniquely. Therefore, raw genomic datasets cannot be publicly published or shared with researchers. The recent success of deep learning (DL) in diverse problems proved its suitability for analyzing the high volume of high-dimensional genomic data. Still, DL-based models leak information about the training samples. To overcome this challenge, we can incorporate differential privacy mechanisms into the DL analysis framework as differential privacy can protect individuals’ privacy. We proposed a differential privacy based DL framework to solve two biological problems: breast cancer status (BCS) and cancer type (CT) classification, and drug sensitivity prediction. To predict BCS and CT using genomic data, we built a differential private (DP) deep autoencoder (dpAE) using private gene expression datasets that performs low-dimensional data representation learning. We used dpAE features to build multiple DP binary classifiers to predict BCS and CT in any individual. To predict drug sensitivity, we used the Genomics of Drug Sensitivity in Cancer (GDSC) dataset. We extracted GDSC’s dpAE features to build our DP drug sensitivity prediction model for 265 drugs. Evaluation of our proposed DP framework shows that it achieves improved prediction performance in predicting BCS, CT, and drug sensitivity than the previously published DP work.

Identification of key apoptosis-related genes and immune infiltration in the pathogenesis of psoriasis

BackgroundPsoriasis is a condition in which skin cells build up and form itchy scales and dry patches. It is also considered a common lifelong disease with an unclear pathogenesis. Furthermore, an effective cure for psoriasis is still unavailable. Reductive apoptosis of keratinocytes and immune infiltration are common in psoriasis. This study aimed to explore underlying functions of key apoptosis-related genes and the characteristics of immune infiltration in psoriasis. We used GSE13355 and GSE30999 to screen differentially expressed apoptosis related genes (DEARGs) in our study. Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway, and gene set enrichment analysis (GSEA) were performed using clusterProfiler package. Protein–protein interaction (PPI) network was constructed to acquire key DEARGs. Transcription factor (TF)–target and miRNA–mRNA network analyses, drug sensitivity prediction, and immune infiltration were applied. Key DEARGs were validated using real-time quantitative PCR (RT-qPCR).ResultsWe identified 482 and 32 DEARGs from GSE13355 and GSE30999, respectively. GO analysis showed that DEARGs were commonly enriched in cell chemotaxis, receptor ligand activity, and signaling receptor activator activity. KEGG pathway analysis indicated that viral protein interaction with cytokine and cytokine receptor was maximally enriched pathway. The GSEA analysis of GSE13355 and GSE30999 demonstrated a high consistency degree of enriched pathways. Thirteen key DEARGs with upregulation were obtained in the PPI network. Eleven key DEARGs were confirmed using RT-qPCR. Additionally, 5 TFs and 553 miRNAs were acquired, and three novel drugs were predicted. Moreover, Dendritic.cells.activated exhibited high levels of immune infiltration while Mast.cells.resting showed low levels of immune infiltration in psoriasis groups.ConclusionResults of this study may reveal some insights into the underlying molecular mechanism of psoriasis and provide novel targeted drugs.

Abstract 1927: Transcell: In silico characterization of genomic landscape and cellular responses from gene expressions through a two-step transfer learning

Abstract Gene expression profiling of new or modified cell lines becomes routine today; however, obtaining comprehensive molecular characterization and cellular responses for a variety of cell lines, including those derived from underrepresented groups, is not trivial when resources are minimal. Using gene expression to predict other measurements has been actively explored; however, systematic investigation of its predictive power in various measurements has not been well studied. We present TransCell, a two-step deep transfer learning framework that utilizes the knowledge derived from pan-cancer tumor samples to predict molecular features and responses. Compared to the five state-of-art methods, TransCell has the best performance in predicting metabolite, gene effect score (or genetic dependency), and drug sensitivity, and has comparable performance in predicting mutation, copy number variation, and protein expression. Notably, TransCell improved the performance by over 50% in drug sensitivity prediction and achieved a correlation of 0.7 in gene effect score prediction. Furthermore, predicted drug sensitivities revealed potential repurposing candidates for new 100 pediatric cancer cell lines, and predicted gene effect scores reflected BRAF resistance in melanoma cell lines. Together, TransCell demonstrates its remarkable predictive power that enables in silico molecular characterization of understudied cell lines. Citation Format: Shan-Ju Yeh, Ruoqiao Chen, Jing Xing, Mengying Sun, Ke Liu, Shreya Paithankar, Jiayu Zhou, Bin Chen. Transcell: In silico characterization of genomic landscape and cellular responses from gene expressions through a two-step transfer learning [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1927.

Abstract 4083: Case study of single-cell protein activity-based drug prediction and validation for precision treatment of cholangiocarcinoma

Abstract Cholangiocarcinoma (CCA) is an aggressive biliary adenocarcinoma, with a median survival ranging from 12 to 37 months and no effective treatment options [Dhanasekaran et al. 2013]. Studies of the CCA tumor micro-environment (TME) and intratumoral heterogeneity have been limited, despite clinically significant interaction between tumor and stromal or immune components across other tumor types. Single-cell RNA Sequencing (scRNASeq) has emerged as a valuable technique to characterize the TME. Here we present a case study profiling CCA TME at the resolution of scRNASeq, and the first application of a unique OncoTarget (OTar) and OncoTreat (OTr) approach [Alvarez et al. 2018, Zeleke et al. 2020, Mundi et al. 2021] to predict and identify actionable drug targets at the single-cell level. These are both CLIA-certified algorithms for personalized drug prediction, now adapted for the first time to drug sensitivity prediction at the level of individual tumor cells. scRNASeq from a human CCA sample revealed significant tumor immune infiltration, with T-cells comprising the largest population, and fewer than 10% of cells identified as tumor cells, such that transcriptional profiles from RNA-Seq derived from whole tumor samples would be dominated by non-tumor cells. VIPER-based protein activity inference of tumor cells identified three sub-populations, not distinguished by gene expression alone [Alvarez et al. 2016]. These were characterized by upregulation of KRAS pathway, TNFa signaling with epithelial-mesenchymal transition, and upregulation of MYC targets, respectively. Consensus OTar/OTr drug prediction analysis on both scRNASeq tumor cells and bulk RNA-Seq of an engrafted PDX model from resected tumor at the time of biopsy ranked Glasdegib, Plicamycin, Flavopiridol, AT9283, and Dacinostat as the top 5 drugs with best overall tumor cell coverage. Therefore, we administered these drugs to a cohort of 8 PDX-bearing mice per treatment arm. Dacinostat and Plicamycin significantly reduced tumor growth rate (p=0.007 and p=0.03, respectively), with Dacinostat stabilizing tumor size over 28 days of treatment. Both of these drugs significantly extended survival time by Kaplan-Meier regression (p=0.001 and p=0.03, respectively). Furthermore, scRNASeq data of drug-treated PDXs showed that Dacinostat uniformly depleted all three tumor sub-populations compared to Vehicle control, whereas one of the tumor sub-clusters was resistant to Plicamycin, consistent with single-cell drug sensitivity predicted by OTr. Given the in vivo activity of these two drugs in inhibiting tumor growth, and the effectiveness of Dacinostat across observed tumor cell phenotypes, as well as the high immune infiltration of this CCA sample, these drugs may be translated into pre-clinical and clinical trials alone and in combination with checkpoint immunotherapy. Citation Format: Lorenzo Tomassoni, Aleksandar Obradovic, Filemon Dela Cruz, Daoqi Yu, Elise Fraser, Susan E. Bates, Yvonne Saenger, Andrea Califano. Case study of single-cell protein activity-based drug prediction and validation for precision treatment of cholangiocarcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 4083.

Abstract 1905: Accelerating clinically-translatable discoveries using a network- and RNA-based precision-oncology framework

Abstract Despite recent advances, prioritizing therapy at the individual patient level remains challenging. In fact, inter-patient tumor heterogeneity remains one of the major challenges in cancer therapy, making it difficult to optimize available treatments on an individual patient basis. Likewise, the systematic prediction of drug sensitivity in vivo is still a major challenge in translational research, where targeted therapeutics are currently selected based on the presence of either actionable oncogene dependencies or aberrant cellular mechanisms. A further challenge is the limited availability of models that faithfully recapitulate the biology, complexity, and heterogeneity of human tumors, including their interaction with a conserved microenvironment and a competent immune system. To address these challenges, we introduce OncoLoop, a highly-generalizable, network-based precision medicine framework to triangulate between available mouse models, human tumors, and large-scale drug perturbational assays with in vivo validation to predict personalized treatment. OncoLoop requires only transcriptomic data (i.e., RNA-seq profiles) and leverages regulatory network analysis to (a) identify cognate models based on conservation of patient-specific Master Regulator (MR) proteins and (b) prioritize drugs based on their ability to invert the activity of MR proteins (MR-inverters), using drug perturbation profiles in cognate cell lines. As proof-of-concept, we applied OncoLoop to prostate cancer using a series of genetically engineered mouse models (GEMMs) that capture a broad range of phenotypes, including metastatic, castration-resistant and neuroendocrine disease. Indeed, >70% of patients in published cohorts had at least one high-fidelity matched GEMM. Drugs targeting shared Master Regulator dependencies of a patient and its cognate GEMM(s) were predicted using perturbational profiles of >300 drugs in MR-matched cell lines, resulting in an 80% validation rate in GEMM allografts and human xenografts. This network-based approach is highly generalized and can be applied to both cancer and non-cancer-related contexts. Citation Format: Alessandro Vasciaveo, Min Zou, Juan M. Arriaga, Francisca Nunes de Almeida, Eugene F. Douglass, Michael Shen, Andrea Califano, Cory Abate-Shen. Accelerating clinically-translatable discoveries using a network- and RNA-based precision-oncology framework [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1905.

Comprehensive Analysis and Reinforcement Learning of Hypoxic Genes Based on Four Machine Learning Algorithms for Estimating the Immune Landscape, Clinical Outcomes, and Therapeutic Implications in Patients With Lung Adenocarcinoma

BackgroundPatients with lung adenocarcinoma (LUAD) exhibit significant heterogeneity in therapeutic responses and overall survival (OS). In recent years, accumulating research has uncovered the critical roles of hypoxia in a variety of solid tumors, but its role in LUAD is not currently fully elucidated. This study aims to discover novel insights into the mechanistic and therapeutic implications of the hypoxia genes in LUAD cancers by exploring the potential association between hypoxia and LUAD.MethodsFour machine learning approaches were implemented to screen out potential hypoxia-related genes for the prognosis of LUAD based on gene expression profile of LUAD samples obtained from The Cancer Genome Atlas (TCGA), then validated by six cohorts of validation datasets. The risk score derived from the hypoxia-related genes was proven to be an independent factor by using the univariate and multivariate Cox regression analyses and Kaplan–Meier survival analyses. Hypoxia-related mechanisms based on tumor mutational burden (TMB), the immune activity, and therapeutic value were also performed to adequately dig deeper into the clinical value of hypoxia-related genes. Finally, the expression level of hypoxia genes was validated at protein level and clinical samples from LUAD patients at transcript levels.ResultsAll patients in TCGA and GEO-LUAD group were distinctly stratified into low- and high-risk groups based on the risk score. Survival analyses demonstrated that our risk score could serve as a powerful and independent risk factor for OS, and the nomogram also exhibited high accuracy. LUAD patients in high-risk group presented worse OS, lower TMB, and lower immune activity. We found that the model is highly sensitive to immune features. Moreover, we revealed that the hypoxia-related genes had potential therapeutic value for LUAD patients based on the drug sensitivity and chemotherapeutic response prediction. The protein and gene expression levels of 10 selected hypoxia gene also showed significant difference between LUAD tumors tissues and normal tissues. The validation experiment showed that the gene transcript levels of most of their genes were consistent with the levels of their translated proteins.ConclusionsOur study might contribute to the optimization of risk stratification for survival and personalized management of LUAD patients by using the hypoxia genes, which will provide a valuable resource that will guide both mechanistic and therapeutic implications of the hypoxia genes in LUAD cancers.

Integrated Analysis of ECT2 and COL17A1 as Potential Biomarkers for Pancreatic Cancer.

Background Pancreatic cancer (PC) is a malignant tumor of the digestive tract. It presents with atypical clinical symptoms and lacks specific diagnostic indicators. This study is aimed at exploring the potential biomarkers of PC. Methods TCGA database pancreatic cancer dataset was normalized and used to identify differentially expressed genes (DEGs). Survival, independent prognostic, and clinical correlation analyses were performed on DEGs to screen for key genes. DNA methylation, mutation, and copy number variation (CNV) analyses were used to analyze genetic variants in key genes. GSEA was performed to explore the functional enrichment of the key genes. Based on the expression of key genes, construction of a competing endogenous RNA (ceRNA) network, analysis of the tumor microenvironment (TME), and prediction of chemotherapeutic drug sensitivity were performed. Furthermore, the GEO database was used to validate the reliability of key genes. Results Two key genes (ECT2 and COL17A1) were identified, which were highly expressed in PC. The mRNA expression of ECT2 and COL17A1 was associated with DNA methylation and CNV. The cell cycle, proteasome, and pathways in cancer were enriched in the high-COL17A1 and ECT2 groups. The TME results showed that immune scores were decreased in the high-ECT2 group. CeRNA network results showed that there were eleven miRNAs were involved in the regulation of ECT2 and COL17A1. Moreover, pRRophetic analysis showed that 20 chemotherapeutic drugs were associated with ECT2 and COL17A1 expression. Conclusions Collectively, ECT2 and COL17A1 may be potential biomarkers for PC, providing a new direction for clinical diagnosis and treatment.

Multi-Omics Integrative Analysis of Lung Adenocarcinoma: An in silico Profiling for Precise Medicine.

Lung adenocarcinoma (LUAD) is one of the most common histological subtypes of lung cancer. The aim of this study was to construct consensus clusters based on multi-omics data and multiple algorithms. In order to identify specific molecular characteristics and facilitate the use of precision medicine on patients we used gene expression, DNA methylation, gene mutations, copy number variation data, and clinical data of LUAD patients for clustering. Consensus clusters were obtained using a consensus ensemble of five multi-omics integrative algorithms. Four molecular subtypes were identified. The CS1 and CS2 subtypes had better prognosis. Based on the immune and drug sensitivity predictions, we inferred that CS1 may be less responsive to immunotherapy and less sensitive to chemotherapeutic drugs. The high immune infiltration of CS2 cells may respond well to immunotherapy. Additionally, the CS2 subtype may also respond to EGFR molecular targeted therapy. The CS3 and CS4 subtypes were associated with poor prognosis. These two subtypes had more mutations, especially TP53 ones, as well as higher sensitivity to chemotherapeutics for lung cancer. However, CS3 was enriched in immune-related pathways and may respond to anti-PD1 immunotherapy. In addition, CS1 and CS4 were less sensitive to ferroptosis inhibitors. We performed a comprehensive analysis of the five types of omics data using five clustering algorithms to reveal the molecular characteristics of LUAD patients. These findings provide new insights into LUAD subtypes and potential clinical treatment strategies to guide personalized management and treatment.

Single-Cell Sequencing Reveals that DBI is the Key Gene and Potential Therapeutic Target in Quiescent Bladder Cancer Stem Cells.

Background: Drug resistance and recurrence often develop during the treatment of muscle-invasive bladder cancer (MIBC). The existence of cancer stem cells (CSCs) in MIBC makes the formulation of effective treatment strategies extremely challenging. We aimed to use single-cell RNA sequencing approaches to identify CSCs and evaluate their molecular characteristics and to discover possible therapeutic measures. Methods: GEO data sets GSE130001 and GSE146137 were used to construct an expression matrix. After cells were identified by type, malignant epithelial cells inferred by InferCNV were extracted for stemness evaluation. The subset of cells with the highest stemness was subjected to weighted gene coexpression network analysis (WGCNA) and pseudotime analysis to identify key genes. In addition, we predicted drug sensitivity relationships for key genes in CTD and predicted the correlation between drugs and survival through siGDC. Results: We found that there were some CSCs in MIBC samples. The CSC population was heterogeneous during tumor development and was divided into quiescent and proliferating CSCs. We identified DBI as the key gene in quiescent CSCs. Analysis of a TCGA data set showed that higher DBI expression indicated higher histological grade. In addition, we predicted that acetaminophen can reduce DBI expression, thereby reducing the stemness of CSCs. Thus, we identified a potential new use of acetaminophen. Conclusion: We systematically explored CSCs in tumors and determined that DBI may be a key gene and potential therapeutic target in quiescent CSCs. In addition, we confirmed that acetaminophen may be a candidate drug targeting CSCs, improving our understanding of CSC-targeting therapeutic strategies.

SEMCM: A Self-Expressive Matrix Completion Model for Anti-cancer Drug Sensitivity Prediction

Background: Genomic data sets generated by several recent large scale high-throughput screening efforts pose a complex computational challenge for anticancer drug sensitivity prediction. Objective: We aimed to design an algorithm model that would predict missing elements in incomplete matrices and could be applicable to drug response prediction programs. Method: We developed a novel self-expressive matrix completion model to improve the predictive performance of drug response prediction problems. The model is based on the idea of subspace clustering and as a convex problem, it can be solved by alternating direction method of multipliers. The original incomplete matrix can be filled through model training and parameters updated iteratively. Results: We applied SEMCM to Genomics of Drug Sensitivity in Cancer (GDSC) and Cancer Cell Line Encyclopedia (CCLE) datasets to predict unknown response values. A large number of experiments have proved that the algorithm has good prediction results and stability, which are better than several existing advanced drug sensitivity prediction and matrix completion algorithms. Without modeling mutation information, SEMCM could correctly predict cell line-drug associations for mutated cell lines and wild cell lines. SEMCM can also be used for drug repositioning. The newly predicted drug responses of GDSC dataset suggest that TI-73 was sensitive to Erlotinib. Moreover, the sensitivity of A172 and NCIH1437 to Paclitaxel was roughly the same. Conclusion: We report an efficient anticancer drug sensitivity prediction algorithm which is opensource and can predict the unknown responses of cancer cell lines to drugs. Experimental results prove that our method can not only improve the prediction accuracy but also can be applied to drug repositioning.

Breast Cancer Prognosis Prediction and Immune Pathway Molecular Analysis Based on Mitochondria-Related Genes.

Background Mitochondria play an important role in breast cancer (BRCA). We aimed to build a prognostic model based on mitochondria-related genes. Method Univariate Cox regression analysis, random forest, and the LASSO method were performed in sequence on pretreated TCGA BRCA datasets to screen out genes from a Gene Set Enrichment Analysis, Gene Ontology: biological process gene set to build a prognosis risk score model. Survival analyses and ROC curves were performed to verify the model by using the GSE103091 dataset. The BRCA datasets were equally divided into high- and low-risk score groups. Comparisons between clinical features and immune infiltration related to different risk scores and gene mutation analysis and drug sensitivity prediction were performed for different groups. Result Four genes, MRPL36, FEZ1, BMF, and AFG1L, were screened to construct our risk score model in which the higher the risk score, the poorer the prognosis. Univariate and multivariate analyses showed that the risk score was significantly associated with age, M stage, and N stage. The gene mutation probability in the high-risk score group was significantly higher than that in the low-risk score group. Patients with higher risk scores were more likely to die. Drug sensitivity prediction in different groups indicated that PF-562271 and AS601245 might be new inhibitors of BRCA. Conclusion We developed a new workable risk score model based on mitochondria-related genes for BRCA prognosis and identified new targets and drugs for BRCA research.

BOSO: A novel feature selection algorithm for linear regression with high-dimensional data.

With the frenetic growth of high-dimensional datasets in different biomedical domains, there is an urgent need to develop predictive methods able to deal with this complexity. Feature selection is a relevant strategy in machine learning to address this challenge. We introduce a novel feature selection algorithm for linear regression called BOSO (Bilevel Optimization Selector Operator). We conducted a benchmark of BOSO with key algorithms in the literature, finding a superior accuracy for feature selection in high-dimensional datasets. Proof-of-concept of BOSO for predicting drug sensitivity in cancer is presented. A detailed analysis is carried out for methotrexate, a well-studied drug targeting cancer metabolism.

Novel acetylation-related gene signatures for predicting the prognosis of patients with colorectal cancer.

Histone acetylation may affect the tumorigenesis and prognosis of colorectal cancer (CRC). However, there is still a lack of studies exploring the effect of acetylation-related genes on the prognosis of CRC. To explore the role of acetylation-related genes in CRC prognosis using bioinformatics strategies, the expression data and survival information of CRC patients were collected from the Gene Expression Omnibus. The Molecular Signatures Database was used to select acetylation-related genes. Univariate and least absolute shrinkage and selection operator regression analyses were used to screen prognostic genes. Kaplan-Meier curves were plotted for survival analysis. Cibersort and pRRophetics were used to analyze immune infiltration and predict drug sensitivity, respectively. By implementing independent prognostic factors, a nomogram model was constructed. The result showed that a total of 48 prognostic genes which screened from the acetylation-related gene set were mainly enriched in ABC transporters and acetylation/deacetylation-related pathways. Three gene signatures (SDR16C5, MEAF6, and SOX4) were further defined, and a prognostic model was constructed that showed high sensitivity and specificity for predicting CRC prognosis in both training and validation cohorts. Patients with different prognostic risks also presented differential expression of gene signatures, infiltration of activated CD4 memory T cells, and drug sensitivity to bicalutamide, gefitinib, Lenalidomide, and imatinib. The nomogram suggested the potential of a risk score-based model in predicting 1- and 2-year survival in patients with CRC. In conclusion, we proposed three gene signatures from an acetylation-related gene set as potential targets for epigenetic therapy and constructed a prognostic model for CRC.