Identification Of Driver Genes Research Articles

Assessing the validity of driver gene identification tools for targeted genome sequencing data.

Most cancer driver gene identification tools have been developed for whole-exome sequencing data. Targeted sequencing is a popular alternative to whole-exome sequencing for large cancer studies due to its greater depth at a lower cost per tumor. Unlike whole-exome sequencing, targeted sequencing only enables mutation calling for a selected subset of genes. Whether existing driver gene identification tools remain valid in that context has not previously been studied. We evaluated the validity of seven popular driver gene identification tools when applied to targeted sequencing data. Based on whole-exome data of 14 different cancer types from TCGA, we constructed matching targeted datasets by keeping only the mutations overlapping with the pan-cancer MSK-IMPACT panel and, in the case of breast cancer, also the breast-cancer-specific B-CAST panel. We then compared the driver gene predictions obtained on whole-exome and targeted mutation data for each of the seven tools. Differences in how the tools model background mutation rates were the most important determinant of their validity on targeted sequencing data. Based on our results, we recommend OncodriveFML, OncodriveCLUSTL, 20/20+, dNdSCv, and ActiveDriver for driver gene identification in targeted sequencing data, whereas MutSigCV and DriverML are best avoided in that context. Code for the analyses is available at https://github.com/SchmidtGroupNKI/TGSdrivergene_validity.

Identification of methylation driver genes for predicting the prognosis of pancreatic cancer patients based on whole-genome DNA methylation sequencing technology

This study was based on the use of whole-genome DNA methylation sequencing technology to identify DNA methylation biomarkers in tumor tissue that can predict the prognosis of patients with pancreatic cancer (PCa). TCGA database was used to download PCa-related DNA methylation and transcriptome atlas data. Methylation driver genes (MDGs) were obtained using the MethylMix package. Candidate genes in the MDGs were screened for prognostic relevance to PCa patients by univariate Cox analysis, and a prognostic risk score model was constructed based on the key MDGs. ROC curve analysis was performed to assess the accuracy of the prognostic risk score model. The effects of PIK3C2B knockdown on malignant phenotypes of PCa cells were investigated in vitro. A total of 2737 differentially expressed genes were identified, with 649 upregulated and 2088 downregulated, using 178 PCa samples and 171 normal samples. MethylMix was employed to identify 71 methylation-driven genes (47 hypermethylated and 24 hypomethylated) from 185 TCGA PCa samples. Cox regression analyses identified eight key MDGs (LEF1, ZIC3, VAV3, TBC1D4, FABP4, MAP3K5, PIK3C2B, IGF1R) associated with prognosis in PCa. Seven of them were hypermethylated, while PIK3C2B was hypomethylated. A prognostic risk prediction model was constructed based on the eight key MDGs, which was found to accurately predict the prognosis of PCa patients. In addition, the malignant phenotypes of PANC-1 cells were decreased after the knockdown of PIK3C2B. Therefore, the prognostic risk prediction model based on the eight key MDGs could accurately predict the prognosis of PCa patients.

A feature extraction framework for discovering pan‐cancer driver genes based on multi‐omics data

AbstractThe identification of tumor driver genes facilitates accurate cancer diagnosis and treatment, playing a key role in precision oncology, along with gene signaling, regulation, and their interaction with protein complexes. To tackle the challenge of distinguishing driver genes from a large number of genomic data, we construct a feature extraction framework for discovering pan‐cancer driver genes based on multi‐omics data (mutations, gene expression, copy number variants, and DNA methylation) combined with protein–protein interaction (PPI) networks. Using a network propagation algorithm, we mine functional information among nodes in the PPI network, focusing on genes with weak node information to represent specific cancer information. From these functional features, we extract distribution features of pan‐cancer data, pan‐cancer TOPSIS features of functional features using the ideal solution method, and SetExpan features of pan‐cancer data from the gene functional features, a method to rank pan‐cancer data based on the average inverse rank. These features represent the common message of pan‐cancer. Finally, we use the lightGBM classification algorithm for gene prediction. Experimental results show that our method outperforms existing methods in terms of the area under the check precision‐recall curve (AUPRC) and demonstrates better performance across different PPI networks. This indicates our framework’s effectiveness in predicting potential cancer genes, offering valuable insights for the diagnosis and treatment of tumors.

Abstract 6248: Multilayered molecular interactions underlying glioblastoma pathogenesis through multi-omics analysis

Abstract Glioblastoma, IDH-wild type (GBM) is the most common malignant and treatment-refractory brain tumor with a dismal prognosis. Despite the identification of various driver genes in previous large-scale genomic analyses, there has been limited progress in the development of novel treatments. Acquiring an extensive understanding of the molecular mechanisms underlying GBM pathogenesis is crucial to improve prognostic outcomes. We analyzed 289 whole-genome sequencing data (WGS) including 159 unpublished deep WGS (≥ ×120 coverage) along with RNA-seq, DNA methylation array, and assay for transposase-accessible chromatin with sequencing (ATAC-seq) to uncover the molecular mechanisms regulating development and progression of GBM. Our deep WGS enables us to delineate a fine view of clonal architecture, where mutational signature varies between clonal and subclonal mutations, suggesting that different mutational processes contribute to GBM pathogenesis depending on its developmental stage. The differentiation status of tumor cells detected by transcriptional deconvolution analysis is associated with multi-layer profiles including genomic alterations, expression subtypes, DNA methylation, and open chromatin status. Tumors predominantly comprised of differentiated cells display genetic and epigenetic profiles that align with the classical subtype, whereas tumors predominantly composed of stem-like cells exhibit profiles consistent with the proneural subtype. ATAC-seq unveils genome-wide chromatin accessibility features associated with gene expression subtypes. Motif enrichment analysis of differentially accessible sites among the subtypes identified specific transcription factors. The proneural subtype is enriched for the SOX10 binding motif, which is associated with regulating cell states, and exhibits higher expression of SOX10 compared to other subtypes. Conversely, the mesenchymal and classical subtypes are enriched for the CREB1 binding motif, which promotes cell proliferation and angiogenesis through TGFβ2 upregulation, wherein TGFβ2 expression is significantly elevated. These findings support a model in which the difference in chromatin structure also regulates the progression of GBM.Our analysis reveals the multi-layer molecular interactions underlying the progression of GBM enhancing our understanding of the pathogenesis. Citation Format: Takuma Nakashima, Yusuke Funakoshi, Ryo Yamamoto, Yuriko Sugihara, Shohei Nambu, Yoshiki Arakawa, Shota Tanaka, Joji Ishida, Ryuta Saito, Ryosuke Hanaya, Koji Yoshimoto, Yoshitaka Narita, Hiromichi Suzuki. Multilayered molecular interactions underlying glioblastoma pathogenesis through multi-omics analysis [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 6248.

Abstract 1611: DNA methylation co-operates with genomic alterations during non-small cell lung cancer evolution

Abstract Aberrant DNA methylation patterns have been described in nearly all human cancers, yet their interplay with genomic alterations during tumor evolution is poorly understood. To explore this crosstalk, we performed reduced representation bisulfite sequencing on 217 tumor regions from 59 patients from the TRACERx study, all with matched adjacent normal tissue, to deconvolve tumor methylation rates. We combined two new metrics for integrative evolutionary analysis with DNA and RNA sequencing data. I-TMD (intra-tumoral methylation distance) quantifies intra-tumor DNA methylation heterogeneity. MR/MN classifies genes based on the rate of hypermethylation at regulatory (MR) versus non-regulatory (MN) CpGs, to identify driver genes exhibiting recurrent functional hypermethylation. We observed early DNA methylation of VIPR2 and ZNF714 co-occurring with driver mutations in CDKN2A and STK11, respectively. Finally, we found DNA methylation-linked dosage compensation of essential genes co-amplified with neighboring oncogenes. Our identification of hypermethylated driver genes under positive selection may open new avenues for therapeutic stratification of patients with non-small cell lung cancer. Citation Format: Francisco Gimeno-Valiente, Carla Castignani, Elizabeth Cadieux L. Cadieux, Nana E. Mensah, Olga Chervova, Mariam Jamal-Hanjani, Nicholas McGranahan, Stephan Beck, Jonas Demeulemeester, Miljana Tanic, Charles Swanton, Peter Van Loo, Nnennya Kanu. DNA methylation co-operates with genomic alterations during non-small cell lung cancer evolution [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 1611.

Identification of cancer driver genes based on hierarchical weak consensus model.

The online version contains supplementary material available at 10.1007/s13755-024-00279-6.

A critical appreciation of pathway analysis in atherosclerotic disease. Cellular phenotypic plasticity as an illustrative example

The rapid advancements in genome-scale (omics) techniques has created significant opportunities to investigate complex disease mechanisms in tissues and cells. Nevertheless, interpreting -omics data can be challenging, and pathway enrichment analysis is a frequently used method to identify candidate molecular pathways that drive gene expression changes. With a growing number of -omics studies dedicated to atherosclerosis, there has been a significant increase in studies and hypotheses relying on enrichment analysis.This brief review discusses the benefits and limitations of pathway enrichment analysis within atherosclerosis research. We highlight the challenges of identifying complex biological processes, such as cell phenotypic switching, within -omics data. Additionally, we emphasize the need for more comprehensive and curated gene sets that reflect the biological complexity of atherosclerosis.Pathway enrichment analysis is a valuable tool for gaining insights into the molecular mechanisms of atherosclerosis. Nevertheless, it is crucial to remain aware of the intrinsic limitations of this approach. By addressing these weaknesses, enrichment analysis in atherosclerosis can lead to breakthroughs in identifying the mechanisms of disease progresses, the identification of key driver genes, and consequently, advance personalized patient care.

Privacy-Preserving Identification of Cancer Subtype-Specific Driver Genes Based on Multigenomics Data with Privatedriver.

Identifying cancer subtype-specific driver genes from a large number of irrelevant passengers is crucial for targeted therapy in cancer treatment. Recently, the rapid accumulation of large-scale cancer genomics data from multiple institutions has presented remarkable opportunities for identification of cancer subtype-specific driver genes. However, the insufficient subtype samples, privacy issues, and heterogenous of aberration events pose great challenges in precisely identifying cancer subtype-specific driver genes. To address this, we introduce privatedriver, the first model for identifying subtype-specific driver genes that integrates genomics data from multiple institutions in a data privacy-preserving collaboration manner. The process of identifying subtype-specific cancer driver genes using privatedriver involves the following two steps: genomics data integration and collaborative training. In the integration process, the aberration events from multiple genomics data sources are combined for each institution using the forward and backward propagation method of NetICS. In the collaborative training process, each institution utilizes the federated learning framework to upload encrypted model parameters instead of raw data of all institutions to train a global model by using the non-negative matrix factorization algorithm. We applied privatedriver on head and neck squamous cell and colon cancer from The Cancer Genome Atlas website and evaluated it with two benchmarks using macro-Fscore. The comparison analysis demonstrates that privatedriver achieves comparable results to centralized learning models and outperforms most other nonprivacy preserving models, all while ensuring the confidentiality of patient information. We also demonstrate that, for varying predicted driver gene distributions in subtype, our model fully considers the heterogeneity of subtype and identifies subtype-specific driver genes corresponding to the given prognosis and therapeutic effect. The success of privatedriver reveals the feasibility and effectiveness of identifying cancer subtype-specific driver genes in a data protection manner, providing new insights for future privacy-preserving driver gene identification studies.

Multi-Network Graph Contrastive Learning for cancer driver gene identification

Multi-Network Graph Contrastive Learning for cancer driver gene identification

A Review of Recent Advances in the Molecular Mechanisms Underlying Brain Metastasis in Lung Cancer.

Brain metastasis from lung cancer is a prevalent mode of treatment failure associated with a poor prognosis. The incidence of brain metastasis has recently shown a dramatic increase. The early detection and risk stratification of lung cancer-related brain metastasis would be highly advantageous for patients. However, our current knowledge and comprehension of the underlying mechanisms driving brain metastasis in lung cancer pose significant challenges. This review summarizes the mechanisms underlying brain metastasis, focusing on the intricate interplay between lung cancer-derived tumor cells and the unique characteristics of the brain, recent advancements in the identification of driver genes, concomitant genes, epigenetic features, including miRNAs and long noncoding RNAs, as well as the molecular characterization of brain metastasis originating from other organs, which may further enhance risk stratification and facilitate precise treatment strategies.

Identification of driver genes in lupus nephritis based on comprehensive bioinformatics and machine learning.

Lupus nephritis (LN) is a common and severe glomerulonephritis that often occurs as an organ manifestation of systemic lupus erythematosus (SLE). However, the complex pathological mechanisms associated with LN have hindered the progress of targeted therapies. We analyzed glomerular tissues from 133 patients with LN and 51 normal controls using data obtained from the GEO database. Differentially expressed genes (DEGs) were identified and subjected to enrichment analysis. Weighted gene co-expression network analysis (WGCNA) was utilized to identify key gene modules. The least absolute shrinkage and selection operator (LASSO) and random forest were used to identify hub genes. We also analyzed immune cell infiltration using CIBERSORT. Additionally, we investigated the relationships between hub genes and clinicopathological features, as well as examined the distribution and expression of hub genes in the kidney. A total of 270 DEGs were identified in LN. Using weighted gene co-expression network analysis (WGCNA), we clustered these DEGs into 14 modules. Among them, the turquoise module displayed a significant correlation with LN (cor=0.88, p<0.0001). Machine learning techniques identified four hub genes, namely CD53 (AUC=0.995), TGFBI (AUC=0.997), MS4A6A (AUC=0.994), and HERC6 (AUC=0.999), which are involved in inflammation response and immune activation. CIBERSORT analysis suggested that these hub genes may contribute to immune cell infiltration. Furthermore, these hub genes exhibited strong correlations with the classification, renal function, and proteinuria of LN. Interestingly, the highest hub gene expression score was observed in macrophages. CD53, TGFBI, MS4A6A, and HERC6 have emerged as promising candidate driver genes for LN. These hub genes hold the potential to offer valuable insights into the molecular diagnosis and treatment of LN.

Identification of Pancreatic Cancer Driver Genes: Random Forest Modeling of Principal Components of scRNA-seq

Identification of Pancreatic Cancer Driver Genes: Random Forest Modeling of Principal Components of scRNA-seq

SMG: self-supervised masked graph learning for cancer gene identification.

Cancer genomics is dedicated to elucidating the genes and pathways that contribute to cancer progression and development. Identifying cancer genes (CGs) associated with the initiation and progression of cancer is critical for characterization of molecular-level mechanism in cancer research. In recent years, the growing availability of high-throughput molecular data and advancements in deep learning technologies has enabled the modelling of complex interactions and topological information within genomic data. Nevertheless, because of the limited labelled data, pinpointing CGs from a multitude of potential mutations remains an exceptionally challenging task. To address this, we propose a novel deep learning framework, termed self-supervised masked graph learning (SMG), which comprises SMG reconstruction (pretext task) and task-specific fine-tuning (downstream task). In the pretext task, the nodes of multi-omic featured protein-protein interaction (PPI) networks are randomly substituted with a defined mask token. The PPI networks are then reconstructed using the graph neural network (GNN)-based autoencoder, which explores the node correlations in a self-prediction manner. In the downstream tasks, the pre-trained GNN encoder embeds the input networks into feature graphs, whereas a task-specific layer proceeds with the final prediction. To assess the performance of the proposed SMG method, benchmarking experiments are performed on three node-level tasks (identification of CGs, essential genes and healthy driver genes) and one graph-level task (identification of disease subnetwork) across eight PPI networks. Benchmarking experiments and performance comparison with existing state-of-the-art methods demonstrate the superiority of SMG on multi-omic feature engineering.

Identifying Significantly Perturbed Subnetworks in Cancer Using Multiple Protein-Protein Interaction Networks.

The identification of cancer driver genes and key molecular pathways has been the focus of large-scale cancer genome studies. Network-based methods detect significantly perturbed subnetworks as putative cancer pathways by incorporating genomics data with the topological information of PPI networks. However, commonly used PPI networks have distinct topological structures, making the results of the same method vary widely when applied to different networks. Furthermore, emerging context-specific PPI networks often have incomplete topological structures, which pose serious challenges for existing subnetwork detection algorithms. In this paper, we propose a novel method, referred to as MultiFDRnet, to address the above issues. The basic idea is to model a set of PPI networks as a multiplex network to preserve the topological structure of individual networks, while introducing dependencies among them, and, then, to detect significantly perturbed subnetworks on the modeled multiplex network using all the structural information simultaneously. To illustrate the effectiveness of the proposed approach, an extensive benchmark analysis was conducted on both simulated and real cancer data. The experimental results showed that the proposed method is able to detect significantly perturbed subnetworks jointly supported by multiple PPI networks and to identify novel modular structures in context-specific PPI networks.

High-Throughput Identification, Modeling, and Analysis of Cancer Driver Genes In Vivo.

The vast number of genomic and molecular alterations in cancer pose a substantial challenge to uncovering the mechanisms of tumorigenesis and identifying therapeutic targets. High-throughput functional genomic methods in genetically engineered mouse models allow for rapid and systematic investigation of cancer driver genes. In this review, we discuss the basic concepts and tools for multiplexed investigation of functionally important cancer genes in vivo using autochthonous cancer models. Furthermore, we highlight emerging technical advances in the field, potential opportunities for future investigation, and outline a vision for integrating multiplexed genetic perturbations with detailed molecular analyses to advance our understanding of the genetic and molecular basis of cancer.

Identification of Driver Epistatic Gene Pairs Combining Germline and Somatic Mutations in Cancer.

Cancer arises from the complex interplay of various factors. Traditionally, the identification of driver genes focuses primarily on the analysis of somatic mutations. We describe a new method for the detection of driver gene pairs based on an epistasis analysis that considers both germline and somatic variations. Specifically, the identification of significantly mutated gene pairs entails the calculation of a contingency table, wherein one of the co-mutated genes can exhibit a germline variant. By adopting this approach, it is possible to select gene pairs in which the individual genes do not exhibit significant associations with cancer. Finally, a survival analysis is used to select clinically relevant gene pairs. To test the efficacy of the new algorithm, we analyzed the colon adenocarcinoma (COAD) and lung adenocarcinoma (LUAD) samples available at The Cancer Genome Atlas (TCGA). In the analysis of the COAD and LUAD samples, we identify epistatic gene pairs significantly mutated in tumor tissue with respect to normal tissue. We believe that further analysis of the gene pairs detected by our method will unveil new biological insights, enhancing a better description of the cancer mechanism.

A novel heterophilic graph diffusion convolutional network for identifying cancer driver genes.

Identifying cancer driver genes plays a curial role in the development of precision oncology and cancer therapeutics. Although a plethora of methods have been developed to tackle this problem, the complex cancer mechanisms and intricate interactions between genes still make the identification of cancer driver genes challenging. In this work, we propose a novel machine learning method of heterophilic graph diffusion convolutional networks (called HGDCs) to boost cancer-driver gene identification. Specifically, HGDC first introduces graph diffusion to generate an auxiliary network for capturing the structurally similar nodes in a biomolecular network. Then, HGDC designs an improved message aggregation and propagation scheme to adapt to the heterophilic setting of biomolecular networks, alleviating the problem of driver gene features being smoothed by its neighboring dissimilar genes. Finally, HGDC uses a layer-wise attention classifier to predict the probability of one gene being a cancer driver gene. In the comparison experiments with other existing state-of-the-art methods, our HGDC achieves outstanding performance in identifying cancer driver genes. The experimental results demonstrate that HGDC not only effectively identifies well-known driver genes on different networks but also novel candidate cancer genes. Moreover, HGDC can effectively prioritize cancer driver genes for individual patients. Particularly, HGDC can identify patient-specific additional driver genes, which work together with the well-known driver genes to cooperatively promote tumorigenesis.

Abstract 185: Global mapping of pathway modules and phosphorylation networks in PDX and corresponding organoid (PDXO) models treated with targeted therapies

Abstract Introduction: Multi-omics-based strategies in precision medicine, including genomic, transcriptomic and proteomic data, have contributed to the molecular-level characterization of cancers and the identification of novel driver genes, as well as a better comprehension of resistance mechanisms. Patient-derived tumor models, including patient-derived xenograft (PDX) and in vitro 3D organoid counterparts (PDXO), are widely recognized as predictive preclinical cancer models closely recapitulating both tumor complexity and patient response and enabling the study of tumor identity for personalized medicine. Taking advantage of Crown Bioscience large PDX collection, with genomic and phenotypic annotation and pharmacological validation, a series of PDXO models have been established and characterized to be used as scalable and high throughput compatible drug screening platforms. Methods: Drug response in matched PDX and PDXO models were investigated on the functional and systems level through deep whole- and phospho-proteomic analysis. For this purpose, analysis and comparisons were made across dose and time series, the effect of several targeted therapies on protein expression and phosphorylation in both PDX and PDXO tumor models. More specifically, a KRAS inhibitor (AMG510), a BCR-ABL TKI (ponatinib) and an EGFR TKI (afatinib) were tested respectively in a lung NSCLC model carrying KRAS G12C mutation, a colorectal model with RET fusion and a lung NSCLC model with EGFR exon 19 deletion. Results: Based on previous data, a specific relationship between area under the curve (AUC) value of organoid drug dose response and in vivo tumor growth has been observed, irrespective of the drug treatment. To interrogate the pathways involved, both PDX and PDXO models were subjected to deep proteome and phospho-proteome profiling using state-of-the-art mass spectrometry-based workflows. An extensive dataset was generated where 16,797 proteins and 12,578 proteins were robustly quantified in PDX and PDXO models respectively. From the same samples, 35,403 phospho-sites were quantified across all PDX samples while 40,754 phospho-sites were quantified across all PDXO models. These data enabled the validation of known pathways (such as RAF/MEK/ERK, PI3K/AKT) and pinpoint new ones modulated through three small molecule inhibitors each targeting a defined genetic background. Conclusion: The predictivity of organoid cultures was demonstrated to model in vivo drug responses and also to serve as a powerful platform to investigate target identification, mechanism of action and resistance mechanism via functional proteome and phospho-proteome analysis. Citation Format: Martin Mehnert, Xiaoxi Xu, Tobias Treiber, Limei Shang, Leilei Chen, Jessie Wang, Marco Tognetti, Christopher Below, Roland Bruderer, Jakob Vowinckel, Binchen Mao, Wubin Qian, Sheng Guo, Ludovic Bourré, Yuehan Feng. Global mapping of pathway modules and phosphorylation networks in PDX and corresponding organoid (PDXO) models treated with targeted therapies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 185.

Identification of Cancer Driver Genes by Integrating Multiomics Data with Graph Neural Networks

Cancer is a heterogeneous disease that is driven by the accumulation of both genetic and nongenetic alterations, so integrating multiomics data and extracting effective information from them is expected to be an effective way to predict cancer driver genes. In this paper, we first generate comprehensive instructive features for each gene from genomic, epigenomic, transcriptomic levels together with protein-protein interaction (PPI)-networks-derived attributes and then propose a novel semisupervised deep graph learning framework GGraphSAGE to predict cancer driver genes according to the impact of the alterations on a biological system. When applied to eight tumor types, experimental results suggest that GGraphSAGE outperforms several state-of-the-art computational methods for driver genes identification. Moreover, it broadens our current understanding of cancer driver genes from multiomics level and identifies driver genes specific to the tumor type rather than pan-cancer. We expect GGraphSAGE to open new avenues in precision medicine and even further predict drivers for other complex diseases.

IMI-driver: integrating multi-level gene networks and multi-omics for cancer driver gene identification

IMI-driver: integrating multi-level gene networks and multi-omics for cancer driver gene identification