Drug Sensitivity Prediction Research Articles

Integrated bulk and single-cell profiling characterize sphingolipid metabolism in pancreatic cancer

BackgroundAbnormal sphingolipid metabolism (SM) is closely linked to the incidence of cancers. However, the role of SM in pancreatic cancer (PC) remains unclear. This study aims to explore the significance of SM in the prognosis, immune microenvironment, and treatment of PC.MethodsSingle-cell and bulk transcriptome data of PC were acquired via TCGA and GEO databases. SM-related genes (SMRGs) were obtained via MSigDB database. Consensus clustering was utilized to construct SM-related molecular subtypes. LASSO and Cox regression were utilized to build SM-related prognostic signature. ESTIMATE and CIBERSORT algorithms were employed to assess the tumour immune microenvironment. OncoPredict package was used to predict drug sensitivity. CCK-8, scratch, and transwell experiments were performed to analyze the function of ANKRD22 in PC cell line PANC-1 and BxPC-3.ResultsA total of 153 SMRGs were acquired, of which 48 were linked to PC patients’ prognosis. Two SM-related subtypes (SMRGcluster A and B) were identified in PC. SMRGcluster A had a poorer outcome and more active SM process compared to SMRGcluster B. Immune analysis revealed that SMRGcluster B had higher immune and stromal scores and CD8 + T cell abundance, while SMRGcluster A had a higher tumour purity score and M0 macrophages and activated dendritic cell abundance. PC with SMRGcluster B was more susceptible to gemcitabine, paclitaxel, and oxaliplatin. Then SM-related prognostic model (including ANLN, ANKRD22, and DKK1) was built, which had a very good predictive performance. Single-cell analysis revealed that in PC microenvironment, macrophages, epithelial cells, and endothelial cells had relatively higher SM activity. ANKRD22, DKK1, and ANLN have relatively higher expression levels in epithelial cells. Cell subpopulations with high expression of ANKRD22, DKK1, and ANLN had more active SM activity. In vitro experiments showed that ANKRD22 knockdown can inhibit the proliferation, migration, and invasion of PC cells.ConclusionThis study revealed the important significance of SM in PC and identified SM-associated molecular subtypes and prognostic model, which provided novel perspectives on the stratification, prognostic prediction, and precision treatment of PC patients.

Machine learning-based identification and validation of immune-related biomarkers for early diagnosis and targeted therapy in diabetic retinopathy

The early diagnosis of diabetic retinopathy (DR) is challenging, highlighting the urgent need to identify new biomarkers. Immune responses play a crucial role in DR, yet there are currently no reports of machine learning (ML) algorithms being utilized for the development of immune-related molecular markers in DR. Based on the datasets GSE102485 and GSE160306, differentially expressed genes (DEGs) were screened using Weighted Gene Co-expression Network Analysis (WGCNA). Five ML algorithms including Bayesian, Learning Vector Quantization (LVQ), Wrapper (Boruta), Random Forest (RF), and Logistic Regression were employed to select immune-related genes associated with DR (DR.Sig). Seven ML algorithms including Naive Bayes (NB), RF, Support Vector Machine (SVM), AdaBoost Classification Trees (AdaBoost), Boosted Logistic Regressions (LogitBoost), K-Nearest Neighbors (KNN), and Cancerclass were utilized to construct a predictive model for DR. The relationship between DR.Sig genes and immune cells was analyzed using single-sample Gene Set Enrichment Analysis (ssGSEA). Additionally, drug sensitivity prediction of DR.Sig genes and molecular docking were performed. Through the utilization of 5 ML algorithms, 6 immune-related biomarkers closely related to the occurrence of DR were identified, including FCGR2B, CSRP1, EDNRA, SDC2, TEK, and CIITA. The DR predictive model constructed based on these 6 DR.Sig genes using the Cancerclass algorithm demonstrated superior predictive performance compared to 4 previously published DR-related biomarkers. In vivo and in vitro experiments also provided strong validation of the expression of the 6 genes in DR. Positive correlations were observed between these genes and 22 types of immune cells. Molecular docking results revealed that CSRP1, EDNRA, and TEK exhibited the highest affinities with the small molecule compounds etoposide, FR-139317, and camptothecin, respectively. The models constructed based on various ML algorithms can effectively predict the occurrence of DR events and hold potential for targeted drug therapies, providing a basis for the early diagnosis and targeted treatment of DR.

ResisenseNet hybrid neural network model for predicting drug sensitivity and repurposing in breast Cancer

Breast cancer remains a leading cause of mortality among women worldwide, with drug resistance driven by transcription factors and mutations posing significant challenges. To address this, we present ResisenseNet, a predictive model for drug sensitivity and resistance. ResisenseNet integrates transcription factor expression, genomic markers, drugs, and molecular descriptors, employing a hybrid architecture of 1D-CNN + LSTM and DNN to effectively learn long-range and temporal patterns from amino acid sequences and transcription factor data. The model demonstrated exceptional predictive accuracy, achieving a validation accuracy of 0.9794 and a loss value of 0.042. Comprehensive validation included comparisons with state-of-the-art models and ablation studies, confirming the robustness of the developed architecture. ResisenseNet has been applied to repurpose existing anticancer drugs across 14 different cancers, with a focus on breast cancer. Among the malignancies studied, drugs targeting Low-grade Glioma (LGG) and Lung Adenocarcinoma (LUAD) showed increased sensitivity to breast cancer as per ResisenseNet’s assessment. Further evaluation of the predicted sensitive drugs revealed that 14 had no prior history of anticancer activity against breast cancer. These drugs target key signaling pathways involved in breast cancer, presenting novel therapeutic opportunities. ResisenseNet addresses drug resistance by filtering ineffective compounds and enhancing chemotherapy for breast cancer. In vitro studies on sensitive drugs provide valuable insights into breast cancer prognosis, contributing to improved treatment strategies.

Identification and experimental validation of a sialylation–related long noncoding RNA signature for prognosis of bladder cancer

BackgroundThe dysregulation of sialylation plays a pivotal role in cancer progression and metastasis, impacting various aspects of tumor behavior. This study aimed to investigate the prognostic significance of long non-coding RNAs (lncRNAs) in relation to sialylation. Additionally, we aimed to develop a signature of sialylation-related lncRNAs in the context of bladder cancer.MethodsThis study used transcriptomic data and clinical information from the TCGA (the Cancer Genome Atlas) database to screen for sialylation-related lncRNAs and constructed a prognostic model. The relationships between these lncRNAs and biological pathways, immune cell infiltration, drug sensitivity, etc., were analyzed, and the expression of some lncRNAs was validated at the cellular level.ResultsThis study identified 6 prognostic lncRNAs related to sialylation and constructed a risk score model with high predictive accuracy and reliability. The survival period of patients in the high-risk group was significantly lower than that of the low-risk group, and it was related to various biological pathways and immune functions. In addition, this study found differences in the sensitivity of patients in different risk groups to chemotherapy drugs, providing a reference for personalized treatment.ConclusionIn this study, we examined the relationship between sialylation-related lncRNA and the prognosis of bladder cancer, providing new molecular markers and potential targets for diagnosis and treatment. Our research revealed correlations between sialylation-related lncRNA characteristics and clinicopathological features, potential mechanisms, somatic mutations, immune microenvironment, chemotherapy response, and predicted drug sensitivity in bladder cancer. Additionally, in vitro cellular studies were conducted to validate these findings and lay the groundwork for future clinical applications.

Prognostic implications of metabolism-related genes in acute myeloid leukemia.

Acute myeloid leukemia(AML) is a diverse malignancy with a prognosis that varies, being especially unfavorable in older patients and those with high-risk characteristics. Metabolic reprogramming has become a significant factor in AML development , presenting new opportunities for prognostic assessment and therapeutic intervention. Metabolism-related differentially expressed genes (mDEGs) were identified by integrating KEGG metabolic gene lists with AML gene expression data from GSE63270. Using TCGA data, we performed consensus clustering and survival analysis to investigate the prognostic significance of mDEGs. A metabolic risk model was constructed using LASSO Cox reg ression and enhanced by a nomogram incorporated clinical characteristics. The model was validated through receiver operating characteristic (ROC) curves and survival statistics. Gene network analysis was conducted to identify critical prognostic factors. The tumor immune microenvironment was evaluated using CIBERSORT and ESTIMATE algorithms, followed by correlation analysis between immune checkpoint gene expression and risk scores. Drug sensitivity predictions and in vitro assays were performed to explore the effects of mDEGs on cell proliferation and chemoresistance. An 11-gene metabolic prognostic model was established and validated. High-risk patients had worse overall survival in both training and validation cohorts (p < 0.05). The risk score was an independent prognostic factor. High-risk patients showed increased immune cell infiltration and potential response to checkpoint inhibitors but decreased drug sensitivity. The model correlated with sensitivity to drugs such as venetoclax. Carbonic anhydrase 13 (CA13) was identified as a key gene related to prognosis and doxorubicin resistance. Knocking down CA13 reduced proliferation and increased cell death with doxorubicin treatment. A novel metabolic gene signature was developed to stratify risk and predict prognosis in AML, serving as an independent prognostic factor. CA13 was identified as a potential therapeutic target. This study provides new insights into the prognostic and therapeutic implications of metabolic genes in AML.

A Risk Model Developed based on Homologous Recombination Deficiency Genes for Evaluating the Drug Sensitivity and Prognostic Prediction of Lung Adenocarcinoma.

This study was designed to construct a risk model based on homologous recombination deficiency (HRD) to evaluate the prognosis and drug sensitivity for patients with lung adenocarcinoma (LUAD). LUAD is a subtype of lung cancer with unfavorable overall survival (OS) and prognosis. HRD has been widely studied in various tumors, but its role in LUAD has not been fully understood. We aimed to construct an HRD-related risk model for predicting the prognosis and drug sensitivity of patients with LUAD. Gene expression data of the LUAD samples were collected from the Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) database. We extracted HRD genes from previous literature and performed univariate COX analysis to select those closely associated with LUAD prognosis. ConsensusClusterPlus was employed to stratify the samples in the TCGA-LUAD cohort into different subtypes. A RiskScore model was established applying random forest method. Furthermore, immunotherapy response and drug sensitivity were predicted using Tumor Immune Dysfunction and Exclusion (TIDE) software and pRRophytic R package, respectively. Finally, the clinical features between High- and Low- RiskScore groups were compared. A total of 16 HRD genes relevant to LUAD prognosis were selected and used to classify 3 LUAD clusters (C1, C2, and C3). Specifically, C1, with a lower TIDE score displayed higher immune infiltration and immunotherapy benefit and the optimal OS, while C2 was closely correlated with tumor-relevant pathways and had the worst OS. Finally, 4 HRD genes (RAD51AP1, BRCA1, H2AFX, and FANCL) were determined to develop a RiskScore signature. It was found that a higher RiskScore was related to more advanced stages, worse OS, and tumor development pathways. Additionally, the High-RiskScore group with a higher TIDE score was sensitive to 44 traditional chemotherapy drugs. A nomogram combined with RiskScore exhibited an accurate survival prediction ability. The HRD-based RiskScore played a crucial role in LUAD development, showing a strong potential to serve as a prognostic indicator for LUAD. Our findings contributed to the diagnosis of LUAD and its treatment.

Combining clinical and molecular data for personalized treatment in acute myeloid leukemia: A machine learning approach

Background and ObjectiveThe standard of care in Acute Myeloid Leukemia patients has remained essentially unchanged for nearly 40 years. Due to the complicated mutational patterns within and between individual patients and a lack of targeted agents for most mutational events, implementing individualized treatment for AML has proven difficult. We reanalysed the BeatAML dataset employing Machine Learning algorithms. The BeatAML project entails patients extensively characterized at the molecular and clinical levels and linked to drug sensitivity outputs. Our approach capitalizes on the molecular and clinical data provided by the BeatAML dataset to predict the ex vivo drug sensitivity for the 122 drugs evaluated by the project. MethodsWe utilized ElasticNet, which produces fully interpretable models, in combination with a two-step training protocol that allowed us to narrow down computations. We automated the genes’ filtering step by employing two metrics, and we evaluated all possible data combinations to identify the best training configuration settings per drug. ResultsWe report a Pearson correlation across all drugs of 0.36 when clinical and RNA sequencing data were combined, with the best-performing models reaching a Pearson correlation of 0.67. When we trained using the datasets in isolation, we noted that RNA Sequencing data (Pearson: 0.36) attained three times the predictive power of whole exome sequencing data (Pearson: 0.11), with clinical data falling somewhere in between (Pearson 0.26). Lastly, we present a paradigm of clinical significance. We used our models’ prediction as a drug sensitivity score to rank an individual's expected response to treatment. We identified 78 patients out of 89 (88 %) that the proposed drug was more potent than the administered one based on their ex vivo drug sensitivity data. ConclusionsIn conclusion, our reanalysis of the BeatAML dataset using Machine Learning algorithms demonstrates the potential for individualized treatment prediction in Acute Myeloid Leukemia patients, addressing the longstanding challenge of treatment personalization in this disease. By leveraging molecular and clinical data, our approach yields promising correlations between predicted drug sensitivity and actual responses, highlighting a significant step forward in improving therapeutic outcomes for AML patients.

Multi-task deep latent spaces for cancer survival and drug sensitivity prediction.

Cancer is a very heterogeneous disease that can be difficult to treat without addressing the specific mechanisms driving tumour progression in a given patient. High-throughput screening and sequencing data from cancer cell-lines has driven many developments in drug development, however, there are important aspects crucial to precision medicine that are often overlooked, namely the inherent differences between tumours in patients and the cell-lines used to model them in vitro. Recent developments in transfer learning methods for patient and cell-line data have shown progress in translating results from cell-lines to individual patients in silico. However, transfer learning can be forceful and there is a risk that clinically relevant patterns in the omics profiles of patients are lost in the process. We present MODAE, a novel deep learning algorithm to integrate omics profiles from cell-lines and patients for the purposes of exploring precision medicine opportunities. MODAE implements patient survival prediction as an additional task in a drug-sensitivity transfer learning schema and aims to balance autoencoding, domain adaptation, drug-sensitivity prediction, and survival prediction objectives in order to better preserve the heterogeneity between patients that is relevant to survival. While burdened with these additional tasks, MODAE performed on par with baseline survival models, but struggled in the drug-sensitivity prediction task. Nevertheless, these preliminary results were promising and show that MODAE provides a novel AI-based method for prioritizing drug treatments for high-risk patients. https://github.com/UEFBiomedicalInformaticsLab/MODAE.

Prognostic signature and immune landscape of 5-methylcytosine-related long non-coding RNAs in gastric cancer

BackgroundLong non-coding RNAs (lncRNAs) have been demonstrated to be useful in assessing the prognosis of cancer patients. However, few studies have focused on 5-methylcytosine-related lncRNAs (m5C-lncRNAs) in gastric cancer (GC). In this study, we aimed to establish a m5C-lncRNAs prognostic signature (m5C-LPS) and explore its potential impact on guiding clinical practice for GC. MethodsRNA-sequence and clinicopathological data were retrieved from The Cancer Genome Atlas (TCGA) database, while the coexpression of long non-coding RNAs (lncRNAs) was determined using Pearson's correlation analysis. A m5C-LPS model was constructed using univariate and Lasso Cox regression, and its prognostic value and accuracy were subsequently validated. Subsequently, the expression of 11 m5C-lncRNAs was verified via quantitative real-time PCR (qRT-PCR) in gastric cancer (GC) cell lines. The potential biological mechanism of this signature was elucidated using Gene Set Enrichment Analysis (GSEA). Based on the GSEA findings, CIBERSORT and ESTIMATE algorithms were utilized to conduct a comprehensive investigation of the tumor immune microenvironment (TIME) in GC. Additionally, pRRophetic and TIDE algorithms were employed to predict drug sensitivity and the efficacy of immunotherapy for GC patients. Results280 lncRNAs were identified as m5C-lncRNAs, including RHPN1-AS1, AC093752.3, TSC22D1-AS1, AL391152.1, MAGI2-AS3, AC048382.2, AL033527.3, AC007405.2, AC036103.1, CCDC183-AS1, and ADORA2A-AS1. Their prognostic value was validated, and the expression of these 11 lncRNAs was confirmed in four gastric cancer cell lines using quantitative reverse transcription PCR (qRT-PCR). A nomogram incorporating a risk score was developed to provide more precise clinical decision-making. Gene Set Enrichment Analysis (GSEA) showed that many classical signaling pathways related to tumor progression were enriched in this signature. Analyses related to immunity and drug sensitivity demonstrated distinct differences in features between high-risk and low-risk subgroups. ConclusionThe m5C-LPS can predict the survival of gastric cancer (GC) patients, provide novel therapeutic targets, and thus offer more thoughtful perspectives for future clinical decisions regarding GC.

DrugFormer: Graph-Enhanced Language Model to Predict Drug Sensitivity.

Drug resistance poses a crucial challenge in healthcare, with response rates to chemotherapy and targeted therapy remaining low. Individual patient's resistance is exacerbated by the intricate heterogeneity of tumor cells, presenting significant obstacles to effective treatment. To address this challenge, DrugFormer, a novel graph-augmented large language model designed to predict drug resistance at single-cell level is proposed. DrugFormer integrates both serialized gene tokens and gene-based knowledge graphs for the accurate predictions of drug response. After training on comprehensive single-cell data with drug response information, DrugFormer model presents outperformance, with higher F1, precision, and recall in predicting drug response. Based on the scRNA-seq data from refractory multiple myeloma (MM) and acute myeloid leukemia (AML) patients, DrugFormer demonstrates high efficacy in identifying resistant cells and uncovering underlying molecular mechanisms. Through pseudotime trajectory analysisunique drug-resistant cellular states associated with poor patient outcomes are revealed. Furthermore, DrugFormer identifies potential therapeutic targets, such as COX8A, for overcoming drug resistance across different cancer types. In conclusion, DrugFormer represents a significant advancement in the field of drug resistance prediction, offering a powerful tool for unraveling the heterogeneity of cellular response to drugs and guiding personalized treatment strategies.

Prediction of clinical prognosis and drug sensitivity in hepatocellular carcinoma through the combination of multiple cell death pathways.

Hepatocellular carcinoma (HCC) is the sixth most common malignant tumor, highlighting a significant need for reliable predictive models to assess clinical prognosis, disease progression, and drug sensitivity. Recent studies have highlighted the critical role of various programmed cell death pathways, including apoptosis, necroptosis, pyroptosis, ferroptosis, cuproptosis, entotic cell death, NETotic cell death, parthanatos, lysosome-dependent cell death, autophagy-dependent cell death, alkaliptosis, oxeiptosis, and disulfidptosis, in tumor development. Therefore, by investigating these pathways, we aimed to develop a predictive model for HCC prognosis and drug sensitivity. We analyzed transcriptome, single-cell transcriptome, genomic, and clinical information using data from the TCGA-LIHC, GSE14520, GSE45436, and GSE166635 datasets. Machine learning algorithms were used to establish a cell death index (CDI) with seven gene signatures, which was validated across three independent datasets, showing that high CDI correlates with poorer prognosis. Unsupervised clustering revealed three molecular subtypes of HCC with distinct biological processes. Furthermore, a nomogram integrating CDI and clinical information demonstrated good predictive performance. CDI was associated with immune checkpoint genes and tumor microenvironment components using single-cell transcriptome analysis. Drug sensitivity analysis indicated that patients with high CDI may be resistant to oxaliplatin and cisplatin but sensitive to axitinib and sorafenib. In summary, our model offers a precise prediction of clinical outcomes and drug sensitivity for patients with HCC, providing valuable insights for personalized treatment strategies.

Machine learning-based integration develops a mitophagy-related lncRNA signature for predicting the progression of prostate cancer: a bioinformatic analysis

Prostate cancer remains a complex and challenging disease, necessitating innovative approaches for prognosis and therapeutic guidance. This study integrates machine learning techniques to develop a novel mitophagy-related long non-coding RNA (lncRNA) signature for predicting the progression of prostate cancer. Leveraging the TCGA-PRAD dataset, we identify a set of four key lncRNAs and formulate a riskscore, revealing its potential as a prognostic indicator. Subsequent analyses unravel the intricate connections between riskscore, immune cell infiltration, mutational landscapes, and treatment outcomes. Notably, the pan-cancer exploration of YEATS2-AS1 highlights its pervasive impact, demonstrating elevated expression across various malignancies. Furthermore, drug sensitivity predictions based on riskscore guide personalized chemotherapy strategies, with drugs like Carmustine and Entinostat showing distinct suitability for high and low-risk group patients. Regression analysis exposes significant correlations between the mitophagy-related lncRNAs, riskscore, and key mitophagy-related genes. Molecular docking analyses reveal promising interactions between Cyclophosphamide and proteins encoded by these genes, suggesting potential therapeutic avenues. This comprehensive study not only introduces a robust prognostic tool but also provides valuable insights into the molecular intricacies and potential therapeutic interventions in prostate cancer, paving the way for more personalized and effective clinical approaches.

The construction of a prognostic model by apoptosis-related genes to predict survival, immune landscape, and medication in cholangiocarcinoma

BackgroundCholangiocarcinoma (CCA) is a highly aggressive and invasive malignant tumor of the bile duct, with a poor prognosis and a high mortality rate. Currently, there is a lack of effective targeted treatment methods and reliable biomarkers for prognosis. MethodsWe downloaded RNA-seq and clinical data of CCA from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases as training and test sets. The apoptosis-related genes were obtained from the Molecular Signatures Database (MsigDB) database. We used univariate/multivariate Cox regression and Lasso regression analyses to construct a riskscore prognostic model. Based on the median riskscore, we clustered the patients into high-risk (HR) and low-risk (LR) groups. We carried out Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses of differentially expressed genes (DEGs) in HR and LR groups. The single sample gene set enrichment analysis (ssGSEA) was employed to analyze the immune infiltration of the HR and LR groups. The CellMiner database was utilized to predict drugs and perform molecular docking on drugs and target proteins. ResultsWe identified 8 genes with prognostic significance to construct a prognostic model. Results of GO and KEGG demonstrated that DEGs were mainly enriched in biological functions such as fatty acid metabolic processes and pathways such as the cAMP signaling pathway. Results of ssGSEA uncovered that immune cells such as DCs and Macrophages in the HR group, as well as immune functions such as Check-point and Parainflammation, were considerably higher than those in the LR group. Drug sensitivity prediction and results of molecular docking revealed that Rigosertib targeted the prognostic genes MAP3K1. HYPOTHEMYCIN and AMG900 effectively targeted JUN. ConclusionOur project suggested that the prognostic model with apoptotic features can effectively predict prognosis in CCA patients, proffering prognostic biomarkers and potential therapeutic targets for CCA patients.

Trust me if you can: a survey on reliability and interpretability of machine learning approaches for drug sensitivity prediction in cancer.

With the ever-increasing number of artificial intelligence (AI) systems, mitigating risks associated with their use has become one of the most urgent scientific and societal issues. To this end, the European Union passed the EU AI Act, proposing solution strategies that can be summarized under the umbrella term trustworthiness. In anti-cancer drug sensitivity prediction, machine learning (ML) methods are developed for application in medical decision support systems, which require an extraordinary level of trustworthiness. This review offers an overview of the ML landscape of methods for anti-cancer drug sensitivity prediction, including a brief introduction to the four major ML realms (supervised, unsupervised, semi-supervised, and reinforcement learning). In particular, we address the question to what extent trustworthiness-related properties, more specifically, interpretability and reliability, have been incorporated into anti-cancer drug sensitivity prediction methods over the previous decade. In total, we analyzed 36 papers with approaches for anti-cancer drug sensitivity prediction. Our results indicate that the need for reliability has hardly been addressed so far. Interpretability, on the other hand, has often been considered for model development. However, the concept is rather used intuitively, lacking clear definitions. Thus, we propose an easily extensible taxonomy for interpretability, unifying all prevalent connotations explicitly or implicitly used within the field.

Prediction of cancer drug sensitivity based on genomic feature distribution alignment and drug structure information

In recent years, precision medicine has demonstrated wide applications in cancer therapy, and the focus of precision medicine lies in accurately predicting the responses of different patients to drug treatment. We propose a model for predicting cancer drug sensitivity based on genomic feature distribution alignment and drug structure information. This model initially aligns the genomic features from cell lines with those from patients and removes noise from gene expression data. Subsequently, it integrates drug structure features and employs multi-task learning to predict the drug sensitivity of patients. The experimental results on the genomics of drug sensitivity in cancer (GDSC) dataset indicates that this method achieved a reduced mean square error of 0.905 2, an increased correlation coefficient of 0.875 4, and an enhanced accuracy rate of 0.836 0 which significantly outperformed the recently published methods. On the cancer genome atlas (TCGA) dataset, this method demonstrates an improved average recall rate of 0.571 4 and an increased F1-score of 0.658 0 in predicting drug sensitivity, exhibiting excellent generalization performance. The result demonstrates the potential of this method to assist in the selection of clinical treatment plans in the future.

Construction and validation of a prognostic model for bladder cancer based on disulfidptosis-related lncRNAs.

Bladder cancer (BLCA) is a prevalent and aggressive cancer associated with high mortality and poor prognosis. Currently, studies on the role of disulfidptosis-related long non-coding RNAs (DRLs) in BLCA are limited. This study aims to construct a prognostic model based on DRLs to improve the accuracy of survival predictions for patients and identify novel targets for therapeutic intervention in BLCA management. Transcriptomic and clinical datasets for patients with BLCA were obtained from The Cancer Genome Atlas. Using multivariate Cox regression and least absolute shrinkage and selection operator techniques, a risk prognostic signature defined by DRLs was developed. The model's accuracy and prognostic relevance were assessed through Kaplan-Meier survival plots, receiver operating characteristic curves, concordance index, and principal component analysis. Functional and pathway enrichment analyses, including Gene Ontology, Kyoto Encyclopedia of Genes and Genomes, and Gene Set Enrichment Analysis, were conducted to elucidate the underlying biological processes. Immune cell infiltration was quantified using the CIBERSORT algorithm. Differences and functions of immune cells in different risk groups were evaluated through single-sample Gene Set Enrichment Analysis. The Tumor Immune Dysfunction and Exclusion predictor and tumor mutational burden (TMB) assessments were utilized to gauge the likelihood of response to immunotherapy. Drug sensitivity predictions were made using the Genomics of Drug Sensitivity in Cancer database. A robust 8-DRL risk prognostic model, comprising LINC00513, SMARCA5-AS1, MIR4435-2HG, MIR4713HG, AL122035.1, AL359762.3, AC006160.1, and AL590428.1, was identified as an independent prognostic indicator. This model demonstrated strong predictive power for overall survival in patients with BLCA, revealing significant disparities between high- and low-risk groups regarding tumor microenvironment, immune infiltration, immune functions, TMB, Tumor Immune Dysfunction and Exclusion scores, and drug susceptibility. This study introduces an innovative prognostic signature of 8 DRLs, offering a valuable prognostic tool and potential therapeutic targets for bladder carcinoma. The findings have significant implications for TMB, the immune landscape, and patient responsiveness to immunotherapy and targeted treatments.

TransCell: In Silico Characterization of Genomic Landscape and Cellular Responses by Deep Transfer Learning.

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. Here, we evaluated commonly used machine learning methods and presented TransCell, a two-step deep transfer learning framework that utilized the knowledge derived from pan-cancer tumor samples to predict molecular features and responses. Among these models, TransCell had the best performance in predicting metabolite, gene effect score (or genetic dependency), and drug sensitivity, and had 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, we investigated the predictive power of gene expression in six molecular measurement types and developed a web portal (http://apps.octad.org/transcell/) that enables the prediction of 352,000 genomic and cellular response features solely from gene expression profiles.

Constructing an immune-related prognostic signature for predicting prognosis and immune response in hepatocellular carcinoma

BackgroundCurrently, there are few studies on immune-related prognostic analysis of hepatocellular carcinoma (HCC). Our aim was to establish an immune-correlated prognostic model for HCC. MethodsImmune-associated cells were obtained from the scRNA-seq dataset (GSE149614) of HCC. Differentially expressed genes between normal and tumor cells from immune-associated cells and the immune-related genes from the ImmPort database were used to identify immune-related differentially expressed genes (IRDEGs). Subsequently, the risk model was established in the TCGA-LIHC cohort (n = 438) from the Cancer Genome Atlas (TCGA) database by using Kaplan-Meier (K-M) survival curve, univariate/multivariate Cox regression analysis. Subsequently, we further analyzed tumor immune microenvironment characteristics, somatic mutation, immune checkpoint and its ligand expression levels between high- and low-risk groups, as well as drug sensitivity prediction. ICGC cohort was set as the validation cohort. TCGA-LIHC cohort and three independent the Gene Expression Omnibus (GEO) datasets (GSE54236, GSE14520, and GSE64041) was used to verify IRDEGs expression, as well as PCR assays using clinical samples. ResultsThe IRDEGs was composed of 4 genes, namely B2M, SPP1, PPIA, and HRG. The 438 HCC patients were divided into high- and low-risk group. The high-risk group was associated with poor prognosis, including higher T stage, advanced pathological stages, less immune cell infiltration, higher TP53 mutation rate, the high expression of CTLA4 and HAVCR2. Besides, high-risk populations benefit from most chemotherapy drugs. Similarly, the performance of the risk model was validated in the ICGC. All four datasets (TCGA-LIHC cohort, GSE54236, GSE14520, and GSE64041) and clinical q-PCR results demonstrated that, compared with normal samples, the expressions of B2M and HRG were lower in tumor samples, and the expression of SPP1 was higher. ConclusionIn summary, the immune-related prognostic signature had a good predictive performance on prognosis and immunotherapy for HCC patients.

Leveraging independence in high-dimensional mixed linear regression.

We address the challenge of estimating regression coefficients and selecting relevant predictors in the context of mixed linear regression in high dimensions, where the number of predictors greatly exceeds the sample size. Recent advancements in this field have centered on incorporating sparsity-inducing penalties into the expectation-maximization (EM) algorithm, which seeks to maximize the conditional likelihood of the response given the predictors. However, existing procedures often treat predictors as fixed or overlook their inherent variability. In this paper, we leverage the independence between the predictor and the latent indicator variable of mixtures to facilitate efficient computation and also achieve synergistic variable selection across all mixture components. We establish the non-asymptotic convergence rate of the proposed fast group-penalized EM estimator to the true regression parameters. The effectiveness of our method is demonstrated through extensive simulations and an application to the Cancer Cell Line Encyclopedia dataset for the prediction of anticancer drug sensitivity.

HGGN: Prediction of microRNA-Mediated drug sensitivity based on interpretable heterogeneous graph global-attention network

Drug sensitivity significantly influences therapeutic outcomes. Recent discoveries have highlighted the pivotal role of miRNAs in regulating drug sensitivity by modulating genes associated with drug metabolism and action. As crucial regulators of gene expression, miRNAs have emerged as influential factors in determining an individual’s response to pharmaceutical interventions. However, current methods for predicting miRNA-drug sensitivity associations overlook the challenges posed by heterogeneous networks and data sparsity. In this paper, we design a dual-channel feature representation strategy for heterogeneous networks and construct HGGN, an interpretable deep learning framework designed to predict miRNA-drug sensitivity associations. HGGN advances beyond traditional approaches by employing a unique dual-channel feature extraction method for miRNAs and drugs, enhancing information retrieval from miRNA-drug networks. It incorporates a Global Attention mechanism to overcome feature propagation interruption in sparse networks. Our experiments on public datasets demonstrate HGGN’s superior prediction accuracy over existing methods. The AUC and AUPR metrics of HGGN reached 0.9649 and 0.9610 respectively, and the Accuracy, Precision, Recall and F1 metrics were all above 0.9. We constructed the model toward different negative sample selection strategies with an accuracy gap of less than 1%, which proves the robustness of HGGN. Additionally, HGGN’s application in modeling analyses and case studies reveals hidden miRNA-mediated drug sensitivity pathways, showcasing its potential for exploratory analysis.