Omics Features Research Articles

Self-supervised graph representation learning integrates multiple molecular networks and decodes gene-disease relationships

Leveraging molecular networks to discover disease-relevant modules is a long-standing challenge. With the accumulation of interactomes, there is a pressing need for powerful computational approaches to handle the inevitable noise and context-specific nature of biological networks. Here, we introduce Graphene, a two-step self-supervised representation learning framework tailored to concisely integrate multiple molecular networks and adapted to gene functional analysis via downstream re-training. In practice, we first leverage GNN (graph neural network) pre-training techniques to obtain initial node embeddings followed by re-training Graphene using a graph attention architecture, achieving superior performance over competing methods for pathway gene recovery, disease gene reprioritization, and comorbidity prediction. Graphene successfully recapitulates tissue-specific gene expression across disease spectrum and demonstrates shared heritability of common mental disorders. Graphene can be updated with new interactomes or other omics features. Graphene holds promise to decipher gene function under network context and refine GWAS (genome-wide association study) hits and offers mechanistic insights via decoding diseases from genome to networks to phenotypes.

Deep learning and multi-omics approach to predict drug responses in cancer

BackgroundCancers are genetically heterogeneous, so anticancer drugs show varying degrees of effectiveness on patients due to their differing genetic profiles. Knowing patient’s responses to numerous cancer drugs are needed for personalized treatment for cancer. By using molecular profiles of cancer cell lines available from Cancer Cell Line Encyclopedia (CCLE) and anticancer drug responses available in the Genomics of Drug Sensitivity in Cancer (GDSC), we will build computational models to predict anticancer drug responses from molecular features.ResultsWe propose a novel deep neural network model that integrates multi-omics data available as gene expressions, copy number variations, gene mutations, reverse phase protein array expressions, and metabolomics expressions, in order to predict cellular responses to known anti-cancer drugs. We employ a novel graph embedding layer that incorporates interactome data as prior information for prediction. Moreover, we propose a novel attention layer that effectively combines different omics features, taking their interactions into account. The network outperformed feedforward neural networks and reported 0.90 for R^2 values for prediction of drug responses from cancer cell lines data available in CCLE and GDSC.ConclusionThe outstanding results of our experiments demonstrate that the proposed method is capable of capturing the interactions of genes and proteins, and integrating multi-omics features effectively. Furthermore, both the results of ablation studies and the investigations of the attention layer imply that gene mutation has a greater influence on the prediction of drug responses than other omics data types. Therefore, we conclude that our approach can not only predict the anti-cancer drug response precisely but also provides insights into reaction mechanisms of cancer cell lines and drugs as well.

Multi-omics data fusion using adaptive GTO guided Non-negative matrix factorization for cancer subtype discovery

Background and objective: Cancer subtype discovery is essential for personalized clinical treatment. With the onset of progressive profile techniques for cancer, a large amount of heterogeneous and high-dimensional transcriptomic, proteomic and genomic datasets are easily accumulated. Integrative clustering of such multi-omics data is crucial to recognize their latent structure and to acknowledge the correlation within and across them. Although the integrative analysis of diversified multi-omics data is informative, it is challenging when multiplicity in data inflicts poor accordance w.r.t. clustering structure. The objective of this work is to develop an effective integrative analysis framework that encapsulates the heterogeneity of various biological mechanisms and predicts homogeneous subgroups of cancer patients. Method: In this paper, improved sparse-joint non-negative matrix factorization (sparse-jNMF) has been devised for the problem of cancer-subtype discovery. The initial points of sparse-jNMF have improved using a novel meta-heuristic algorithm adaptive gorilla troops optimizer (Ada-GTO). Improving the initialization of sparse-jNMF enhances its convergence behavior and further strengthens the clustering performance. In addition, the consensus clustering approach has been adopted to construct a patient-patient similarity matrix for obtaining stable clusters of patient samples. Result: The proposed framework has been applied to 4 different real-life multi-omics cancer datasets, namely colon adenocarcinoma, breast-invasive carcinoma, kidney-renal clear-cell carcinoma, and lung adenocarcinoma. The proposed method results in patient clusters with better silhouette scores and cluster purity than classical initialization and similar meta-heuristics for initial point estimation approaches. Survival probabilities estimated using Kaplan-Meier (KM) curve show statistically significant difference (p < 0.05) for the homogenous cancer patient clusters obtained using the proposed method as compared to iCluster. The presented approach further identified the somatic mutations for the classified subgroups, which is beneficial to provide targeted treatments. Conclusion: This paper proposes Ada-GTO guided sparse-jNMF framework for cancer subtype discovery, considering the information from multiple omic features that provide comprehension. The proposed meta-guided framework outperforms all other state-of-the-art counterparts. It also has great potential for obtaining the homogeneous subgroups of other diseases.

SOJNMF: Identifying Multidimensional Molecular Regulatory Modules by Sparse Orthogonality-Regularized Joint Non-Negative Matrix Factorization Algorithm.

Cancer is not only a very aggressive but also a very diverse disease. Recent advances in high-throughput omics technologies of cancer have enabled biomedical researchers to have more opportunities for studying its multi-level biological regulatory mechanism. However, there are few methods to explore the underlying mechanism of cancer by identifying its multidimensional molecular regulatory modules from the multidimensional omics data of cancer. In this paper, we propose a sparse orthogonality-regularized joint non-negative matrix factorization (SOJNMF) algorithm which can integratively analyze multidimensional omics data. This method can not only identify multidimensional molecular regulatory modules, but reduce the overlap rate of features among the multidimensional modules while ensuring the sparsity of the coefficient matrix after decomposition. Gene expression data, miRNA expression data and gene methylation data of liver cancer are integratively analyzed based on SOJNMF algorithm. Then, we obtain 238 multidimensional molecular regulatory modules. The results of permutation test indicate that different omics features within these modules are significantly correlated in statistics. Meanwhile, the results of functional enrichment analysis show that these multidimensional modules are significantly related to the underlying mechanism of the occurrence and development of liver cancer.

THE USE OF MULTI-OMICS TO IDENTIFY PREDICTIVE FACTORS FOR FOOD REACTIONS

<h3>Introduction</h3> The pathophysiology of IgE mediated food reactions (FR) is multifactorial. This multi-omic analyses utilized host and microbial factors to better understand the immunological origins of FR's. <h3>Methods</h3> This longitudinal cohort study included 129 term infants, identified as FR (n=27) or non-FR (n=92) using the Infant Feeding Practices-II survey (confirmed with medical records). Saliva was collected at 6 months for multi-omic analysis of cytokines, mRNA, microRNAs, and the microbiome. <h3>Results</h3> None of the salivary cytokines (CXCL10, IL-8, IL-6, IL-18) displayed differences between groups. However, the Th1/Th2 ratio (IL-8/IL-6) was nominally higher (d = 0.56, p = 0.043, FDR = 0.30) among infants with a food reaction. One mRNA (STAT6, p = 0.039, FDR = 0.17, d = 0.045), two miRNAs (let-7e-5p, p = 0.031, FDR = 0.12, d = 0.51; miR-151a-5p, p = 0.041, FDR = 0.13, d = 0.52), and one bacterial phylum (Cyanobacteria, p = 0.024, FDR = 0.19, d = 0.72) displayed nominal differences between 27 infants with food reactions (FR) and 92 peers with no food reactions (NFR). All four omics features were higher in infants with FR. A logistic regression model employing maternal/infant traits and environmental exposures and four multi-omic features (IL-8/IL-6, miR-151a-5p, STAT6, Cyanobacteria) accounted for an additional 31% of the variance between groups (p = 0.15, AUC = 0.83, sensitivity = 97%, specificity = 48%). <h3>Conclusion</h3> Multi-omic analysis of saliva may enhance understanding of key biological factors that contribute towards developing immune mediated food reactions.

Multi-omics studies reveal ameliorating effects of physical exercise on neurodegenerative diseases.

Neurodegenerative diseases such as Alzheimer's disease, Parkinson's disease, amyotrophic lateral sclerosis, and Huntington's disease, are heavy burdens to global health and economic development worldwide. Mounting evidence suggests that exercise, a type of non-invasive intervention, has a positive impact on the life quality of elderly with neurodegenerative diseases. X-omics are powerful tools for mapping global biochemical changes in disease and treatment. Three major databases were searched related to current studies in exercise intervention on neurodegenerative diseases using omics tools, including metabolomics, metagenomics, genomics, transcriptomics, and proteomics. We summarized the omics features and potential mechanisms associated with exercise and neurodegenerative diseases in the current studies. Three main mechanisms by which exercise affects neurodegenerative diseases were summed up, including adult neurogenesis, brain-derived neurotrophic factor (BDNF) signaling, and short-chain fatty acids (SCFAs) metabolism. Overall, there is compelling evidence that exercise intervention is a feasible way of preventing the onset and alleviating the severity of neurodegenerative diseases. These studies highlight the importance of exercise as a complementary approach to the treatment and intervention of neurodegenerative diseases in addition to traditional treatments. More mechanisms on exercise interventions for neurodegenerative diseases, the specification of exercise prescriptions, and differentiated exercise programs should be explored so that they can actually be applied to the clinic.

RSC-based differential model with correlation removal for improving multi-omics clustering

Multi-omics clustering plays an important role in cancer subtyping. However, the data of different kinds of omics are often related, these correlations may reduce the clustering algorithm performance. It is crucial to eliminate the unexpected redundant information caused by these correlations between different omics. We proposed RSC-based differential model with correlation removal for improving multi-omics clustering (RSC-MCR). This method first introduced RSC to calculate the pairwise correlations of all features, and decomposed it to obtain the pairwise correlations of different omics features, thus built the connection between different omics based on the pairwise correlations of different omics features. Then, to remove the redundant correlation, we designed a differential model to calculate the degree of difference between the original feature matrix and the correlation matrix which contained the most relevant information between different omics. We compared the performance of RSC-MCR with decorrelation methods on different clustering methods (CC, FCM, SNF, NMF, LRAcluster). The experimental results on five cancer datasets show the efficiency of the RSC-MCR as well as improvements over other decorrelation methods.

Multi-omic factors associated with future wheezing in infants.

The pathophysiology of wheezing is multifactorial, impacted by medical, demographic, environmental, and immunologic factors. We hypothesized that multi-omic analyses of host and microbial factors in saliva would enhance the ability to identify infants at risk for wheezing. This longitudinal cohort study included 161 term infants. Infants who developed wheezing (n = 27) within 24 months of delivery were identified using the International Study of Asthma and Allergies in Childhood Written Questionnaire and review of the medical record. Standardized surveys were used to assess infant traits and environmental exposures. Saliva was collected for multi-omic assessment of cytokines, microRNAs, mRNAs, and microbiome/virome RNAs. Two infant factors (daycare attendance, family history of asthma) and three salivary "omic" features (miR-26a-5p, Elusimicrobia, Streptococcus phage phiARI0131-1) differed between the two groups (adjusted p < 0.05). miR-26a-5p levels were correlated with Elusimicrobia (R = -0.87, p = 3.7 × 10-31). A model employing the three omic features plus daycare attendance and family asthma history yielded the highest predictive accuracy for future wheezing episodes (AUC = 0.74, 95% CI: 0.703-0.772, 77% sensitivity, 62% specificity). Host-microbiome interactions in saliva may yield pathophysiologic clues about the origins of wheezing and aid identification of infants at risk of future wheezing episodes. Wheezing is multi-factorial, but the relative contributions of infant traits, environment, and underlying biology are poorly understood. This multi-omic study identifies three molecular factors, including salivary microRNAs, microbes, and viral phages associated with increased risk of infant wheezing. Measurement of these molecular factors enhanced predictive accuracy for future wheezing when combined with family asthma history and daycare attendance. Validation of this approach could be used to identify infants at risk for wheezing and guide personalized medical management.

Mediation analysis of multiple mediators with incomplete omics data.

There is an increasing interest in using multiple types of omics features (e.g., DNA sequences, RNA expressions, methylation, protein expressions, and metabolic profiles) to study how the relationships between phenotypes and genotypes may be mediated by other omics markers. Genotypes and phenotypes are typically available for all subjects in genetic studies, but typically, some omics data will be missing for some subjects, due to limitations such as cost and sample quality. In this article, we propose a powerful approach for mediation analysis that accommodates missing data among multiple mediators and allows for various interaction effects. We formulate the relationships among genetic variants, other omics measurements, and phenotypes through linear regression models. We derive the joint likelihood for models with two mediators, accounting for arbitrary patterns of missing values. Utilizing computationally efficient and stable algorithms, we conduct maximum likelihood estimation. Our methods produce unbiased and statistically efficient estimators. We demonstrate the usefulness of our methods through simulation studies and an application to the Metabolic Syndrome in Men study.

Network-based integration of multi-omics data for clinical outcome prediction in neuroblastoma

Multi-omics data are increasingly being gathered for investigations of complex diseases such as cancer. However, high dimensionality, small sample size, and heterogeneity of different omics types pose huge challenges to integrated analysis. In this paper, we evaluate two network-based approaches for integration of multi-omics data in an application of clinical outcome prediction of neuroblastoma. We derive Patient Similarity Networks (PSN) as the first step for individual omics data by computing distances among patients from omics features. The fusion of different omics can be investigated in two ways: the network-level fusion is achieved using Similarity Network Fusion algorithm for fusing the PSNs derived for individual omics types; and the feature-level fusion is achieved by fusing the network features obtained from individual PSNs. We demonstrate our methods on two high-risk neuroblastoma datasets from SEQC project and TARGET project. We propose Deep Neural Network and Machine Learning methods with Recursive Feature Elimination as the predictor of survival status of neuroblastoma patients. Our results indicate that network-level fusion outperformed feature-level fusion for integration of different omics data whereas feature-level fusion is more suitable incorporating different feature types derived from same omics type. We conclude that the network-based methods are capable of handling heterogeneity and high dimensionality well in the integration of multi-omics.

Prediction of pelvic lymph node metastasis in prostate cancer using radiomics based on T2-weighted imaging.

Pelvic lymph node metastasis (PLNM) is an important factor that affects the stage and prognosis of prostate cancer. Invasive extended pelvic lymph node dissection (ePLND) is the most effective method for clinically diagnosing PLNM. Accurate preoperative prediction of PLNM can reduce unnecessary ePLND. This study aims to investigate the clinical value of radiomics nomogram in predicting PLNM of prostate cancer based on T2-weighted imaging (T2WI). Magnetic resonance (MR) data of 71 patients with prostate cancer who underwent ePLND from January 2017 to June 2021 in Peking University First Hospital were collected retrospectively. All patients were assigned into a training set (January 2017 to December 2020, n=56, containing 186 lymph nodes) and a test set (January 2021 to June 2021, n=15, containing 45 lymph nodes) according to the examination time of multiparametric magnetic resonance imaging (mpMRI). Two radiologists matched the dissected lymph nodes on MRI images, and manually annotated the region of interest (ROI). Based on the outlined ROI, 3 metastatic lymph node prediction models were established: Model 1 (only image features of T2WI), Model 2 (radiomics features based on random forest), and Model 3 (combination of the image and radiomics features). A nomogram was also established. The clinicopathologic characteristics of the patients were obtained from the medical records, including age, the Gleason score, the level of prostate-specific antigen (PSA), and clinical and pathological T stage. The preoperative radiological features of the pelvic lymph nodes (LNs) include size of LNs (the short and long diameters) and volume of LNs. Receiver operating characteristic (ROC) curve was used to evaluate the diagnostic efficacy of the 3 models and decision curve analysis (DCA) was used to evaluate the clinical benefits of the models. No significant differences were found between the training set and test set regarding age, Gleason scores, PSA level, and clinical and pathological T stage (all P>0.05). The differences in volume, short diameter and long diameter between metastatic and non-metastatic LNs were statistically significant in both training set and test set (all P<0.05). In multivariate regression analysis, the short diameter and marginal status of LNs were included in Model 1. Eighteen omics features were selected to construct Model 2. The signal distribution of LNs and Rad score were the significant risk factors for predicting metastasis of pelvic LNs in Model 3. The C-index of nomogram based on Model 3 reached 0.964, and the calibration curve showed that the model had high calibration degree. In the test set, the area under the curves of Model 1, 2, and 3 were 0.78, 0.93, and 0.96 respectively, Model 2 and Model 3 showed significantly higher diagnostic efficiency than Model 1 (Model 1 vs Model 2, P=0.019; Model 1 vs Model 3, P=0.020). There was no significant difference in the area under the curve between Model 2 and Model 3 (P=0.649). The DCA results of the 3 models showed that all models obtained higher net benefits than the PLNM-all or PLNM-none protocol in different ranges of threshold probabilities and Model 3 had the highest clinical benefit. The radiomics nomogram based on T2WI shows a good predictive efficacy for preoperative PLNM in patients with prostate cancer, which could be served as an imaging biomarker to optimize decision-making and adjust adjuvant treatments.

Multi-omics molecular biomarkers and database of osteoarthritis.

Osteoarthritis (OA) is the most common form of arthritis in the adult population and is a leading cause of disability. OA-related genetic loci may play an important role in clinical diagnosis and disease progression. With the rapid development of diverse technologies and omics methods, many OA-related public data sets have been accumulated. Here, we retrieved a diverse set of omics experimental results from 159 publications, including genome-wide association study, differentially expressed genes and differential methylation regions, and 2405 classified OA-related gene markers. Meanwhile, based on recent single-cell RNA-seq data from different joints, 5459 cell-type gene markers of joints were collected. The information has been integrated into an online database named OAomics and molecular biomarkers (OAOB). The database (http://ibi.zju.edu.cn/oaobdb/) provides a web server for OA marker genes, omics features and so on. To our knowledge, this is the first database of molecular biomarkers for OA.

Robust identification of temporal biomarkers in longitudinal omics studies.

MotivationLongitudinal studies increasingly collect rich ‘omics’ data sampled frequently over time and across large cohorts to capture dynamic health fluctuations and disease transitions. However, the generation of longitudinal omics data has preceded the development of analysis tools that can efficiently extract insights from such data. In particular, there is a need for statistical frameworks that can identify not only which omics features are differentially regulated between groups but also over what time intervals. Additionally, longitudinal omics data may have inconsistencies, including non-uniform sampling intervals, missing data points, subject dropout and differing numbers of samples per subject.ResultsIn this work, we developed OmicsLonDA, a statistical method that provides robust identification of time intervals of temporal omics biomarkers. OmicsLonDA is based on a semi-parametric approach, in which we use smoothing splines to model longitudinal data and infer significant time intervals of omics features based on an empirical distribution constructed through a permutation procedure. We benchmarked OmicsLonDA on five simulated datasets with diverse temporal patterns, and the method showed specificity greater than 0.99 and sensitivity greater than 0.87. Applying OmicsLonDA to the iPOP cohort revealed temporal patterns of genes, proteins, metabolites and microbes that are differentially regulated in male versus female subjects following a respiratory infection. In addition, we applied OmicsLonDA to a longitudinal multi-omics dataset of pregnant women with and without preeclampsia, and OmicsLonDA identified potential lipid markers that are temporally significantly different between the two groups.Availability and implementationWe provide an open-source R package (https://bioconductor.org/packages/OmicsLonDA), to enable widespread use.Supplementary information Supplementary data are available at Bioinformatics online.

Abstract 5018: A comprehensive analysis of genomic determinants of response to immune checkpoint inhibitor-based immunotherapy

Abstract A low response rate is the major challenge for existing cancer immunotherapies. A number of biomarkers have been reported to be associated with the likelihood of patient responses to immune checkpoint inhibitors (ICI), including the expression of PD-L1, tumor mutational burden (TMB), tumor infiltration lymphocytes, and T cell repertoire. However, these biomarkers showed variable performance even conflicting conclusions in different cohorts. And a large proportion of patients exhibited an immune-exclusion or immune-desert phenotype that cannot be explained by current biomarkers. In addition, establishing a cohort with adequate sample size requires multiple years of multi-center efforts. Therefore, it is still a challenge and a lack for the evaluation of efficacy of known biomarkers and the discovery of new signature(s) in a large-scale study. We hypothesized that a data-driven meta-analysis approach integrating multi-cohorts with multi-omics data will help to characterize the predictive genomic features and discover the mechanisms of antitumor immune responses. In this work, we collected 3,037 ICI therapy samples with genetic and transcriptomics data for 14 cancer types. Then a data-driven strategy was designed to explore each type of genomic feature which included gene mutations, pathway mutations, TMB, mutational signatures, gene expression, pathway expression, interaction among checkpoint blockade, and immune cell components. Then a comprehensive analysis, which contained a meta-analysis for the identification of consistent biomarkers across multiple cohorts and a data aggregation-analysis for the identification of rare biomarkers on aggregated data, was performed to explore the associations between ICI response and omics features. Furthermore, we validated these signatures as biomarkers in three independent cohorts. Finally, we identified five types of biomarkers, including 1) two biomarkers regarding source of antigen (TMB and mutations of MYO7B), 2) one feature of responding to antigen (cytotoxic T lymphocytes pathway), 3) three biomarkers indicating macrophages M1 infiltrations (macrophages M1 component, and two marker genes of macrophages M1, CXCL11 and CCL19), 4) one biomarker regarding immune interactions (expression relations among immune checkpoint genes), and 5) two biomarkers about T cell-inflamed tumor microenvironment (IFNG and ligands genes of chemokine receptor CXCR3). In addition, a predictive module was built based on these signatures and outperformed than TMB or PD-L1 expression which were approved molecular biomarkers by FDA. We found high TMB patients had more likelihood to active the cytotoxic T lymphocytes which secret IFNG to kill tumor cell and promote the recruitment of macrophage M1. And the T cells were stimulated by macrophage M1 via the binding of CXCR3 of T cell surface and receptor of CXCR3 of macrophage M1 surface. Citation Format: Jing Yang, Qi Liu, Yu Shyr. A comprehensive analysis of genomic determinants of response to immune checkpoint inhibitor-based immunotherapy [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 5018.

Abstract 6341: ImaGene: A robust AI-based software platform for tumor radiogenomic evaluation and reporting

Abstract The field of radiomics has undergone several advancements in approaches to uncovering hidden quantitative features from tumor imaging data for use in guiding clinical decision-making for cancer patients. Radiographic imaging techniques provide insight into the imaging features of tumor regions of interest (ROIs), while immunohistochemistry and sequencing techniques performed on biopsy samples yield omics data. These imaging and omics feature data can then be correlated and modeled using artificial intelligence (AI) techniques to highlight notable associations between tumor genotype and phenotype. Currently, however, the radiogenomics field lacks a unified and robust software platform capable of algorithmically analyzing imaging and omics features using modifiable parameters, detecting significant relationships among these features, and subjecting them to AI-based analysis. To address this gap, we developed ImaGene, a robust AI-based platform that uses omics and imaging features as inputs for different tumor types, performs statistical analyses of the correlations between these data types, and constructs AI models based upon significantly correlated features. It has several modifiable configuration parameters that provide users with complete control over their experiments. For each run, ImaGene produces comprehensive reports that can contribute to the construction of a novel radiogenomic knowledge base, in addition to enabling the deployment and sharing of AI models. To demonstrate the utility of ImaGene, we acquired imaging and omics datasets pertaining to Invasive Breast Cancer (IBC) and Head and Neck Squamous Cell Carcinoma (HNSCC) from public databases and analyzed them with this platform using specific parameters. In both cases, we uncovered significant associations between several imaging features and 11 genes: CRABP1, VRTN, SMTNL2, FABP1, HAND2, HAS1, C4BPA, FAM163A, DSG1, SMTNL2 and KCNJ16 for IBC, and 10 genes: CEACAM6, IGLL1, SERPINA1, NANOG, OCA2, PRLR, ACSM2B, CYP11B1, and VPREB1 for HNSCC. Overall, our software platform is capable of identifying, analyzing, and correlating important features from tumor scans, thereby providing researchers with a reliable and accurate tool for their radiogenomics experiments. We anticipate that ImaGene will become the gold standard for tumor analyses in the field of radiogenomics owing to its ease of use, flexibility, and reproducibility. Citation Format: Shrey S. Sukhadia, Aayush Tyagi, Vivek Venkatraman, Pritam Mukherjee, Prathosh A.P., Mayur Divate, Olivier Gevaert, Shivashankar H. Nagaraj. ImaGene: A robust AI-based software platform for tumor radiogenomic evaluation and reporting [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 6341.

Glycosylation modification identifies novel molecular phenotypes and prognostic stratifications of glioma

Glycosylation modification plays a vital role in tumor progression and is highly associated with glioma prognosis. However, the influence of glycosylation modification on the tumor microenvironment (TME) and omic features of glioma remains unclear. Differentially expressed glycosylation-related genes between adjacent and tumor tissues of The Cancer Genome Atlas and Chinese Glioma Genome Atlas datasets were identified. We performed unsupervised clustering to classify patients into different molecular phenotypes, and analyzed their TME heterogeneity, including immunocyte infiltration, immune pathways and tumor purity. Subsequently, we developed a prognostic predicting system named GlycoScore by stepwise least absolute shrinkage and selection operator-Cox regression to evaluate the modification pattern and its association with somatic mutation, clinical significance, immune fractions and drug resistance. Two clustering clusters were identified and presented distinct clinical outcomes and biological functions characterized by hotand cold tumors respectively. Patients with higher GlycoScores exhibited poor prognosis, less mutation counts, and were more sensitive to chemotherapeutics. We also confirmed that the GlycoScore severed as an independent risk factor. Cancer hallmarks such as cell cycle, hippo, and TGFβ were active in the high-GlycoScore group. The combination of tumor mutation burden and the GlycoScore presented an excellent performance in prognostic stratification. Our study suggests that glycosylation is essential for modeling TME of glioma and the GlycoScore is a promising prognostic signature and indicator of immunotherapeutic efficacy.

Integrated Multi-Omics Maps of Lower-Grade Gliomas

Simple SummaryData from multiple omics domains were increasingly generated in large-scale tumour studies to enhance our understanding of molecular mechanisms of cancer. We present an integrated cartography of three omics layers combining the transcriptome, methylome, and genome (copy number variations) into a unique mapping scheme which enabled us to decipher functional links within and between the omics domains. Application to lower grade gliomas reveals distinct networks governed either by methylation or copy number variations, both affecting transcriptomics modes of cell activity. The integrated maps provide an intuitive view on tumour heterogeneity across the omics layers distinguishing, e.g., astrocytoma- and oligodendroglioma-like glioma types. In a wider sense, multi-omics cartography deciphers the effect of different omes on tumour phenotypes and their molecular hallmarks with individual resolution.Multi-omics high-throughput technologies produce data sets which are not restricted to only one but consist of multiple omics modalities, often as patient-matched tumour specimens. The integrative analysis of these omics modalities is essential to obtain a holistic view on the otherwise fragmented information hidden in this data. We present an intuitive method enabling the combined analysis of multi-omics data based on self-organizing maps machine learning. It “portrays” the expression, methylation and copy number variations (CNV) landscapes of each tumour using the same gene-centred coordinate system. It enables the visual evaluation and direct comparison of the different omics layers on a personalized basis. We applied this combined molecular portrayal to lower grade gliomas, a heterogeneous brain tumour entity. It classifies into a series of molecular subtypes defined by genetic key lesions, which associate with large-scale effects on DNA methylation and gene expression, and in final consequence, drive with cell fate decisions towards oligodendroglioma-, astrocytoma- and glioblastoma-like cancer cell lineages with different prognoses. Consensus modes of concerted changes of expression, methylation and CNV are governed by the degree of co-regulation within and between the omics layers. The method is not restricted to the triple-omics data used here. The similarity landscapes reflect partly independent effects of genetic lesions and DNA methylation with consequences for cancer hallmark characteristics such as proliferation, inflammation and blocked differentiation in a subtype specific fashion. It can be extended to integrate other omics features such as genetic mutation, protein expression data as well as extracting prognostic markers.

Transforming OMIC features for classification using siamese convolutional networks.

Modern biotechnologies have generated huge amount of OMIC data, among which transcriptomes and methylomes are two major OMIC types. Transcriptomes measure the expression levels of all the transcripts while methylomes depict the cytosine methylation levels across a genome. Both OMIC data types could be generated by array or sequencing. And some studies deliver many more features (the number of features is denoted as [Formula: see text]) for a sample than the number [Formula: see text] of samples in a cohort, which induce the "large [Formula: see text] small [Formula: see text]" paradigm. This study focused on the classification problem about OMIC with "large [Formula: see text] small [Formula: see text]" paradigm. A Siamese convolutional network was utilized to transform the OMIC features into a new space with minimized intra-class distances and maximized inter-class distances between the samples. The proposed feature engineering algorithm SiaCo was comprehensively evaluated using both transcriptome and methylome datasets. The experimental data showed that SiaCo generated SiaCo features with improved classification accuracies for binary classification problems, and achieved improvements on the independent test dataset. The individual SiaCo features did not show better inter-class discrimination powers than the original OMIC features. This may be due to that the Siamese convolutional network optimized the collective performances of the SiaCo features, instead of the individual feature's discrimination power. The inherent transformation nature of the Siamese twin network also makes the SiaCo features lack of interpretability. The source code of SiaCo is freely available at http://www.healthinformaticslab.org/supp/resources.php.

A novel molecular subtype of hepatocellular carcinoma based on the tumor purity and tumor microenvironment-related polygenic risk scores.

e15040 Background: The molecular classification of hepatocellular carcinoma (HCC) plays a vital role in the management of precision therapy. Several molecular classification systems, mainly focusing on the tumor cells, were developed according to distinct omics features of HCC. Heterogeneous tumor microenvironment (TME) in HCC is recognized as a crucial part of inter-tumor heterogeneity. Few molecular classifications of HCC have so far taken into account both the related molecules of tumor cells and TME. Methods: Wefirstly identified the tumor purity (TP)-related and TME-related genes in the differentially expressed genes in HCC using GSE14520 data set, and separately constructed the TP-related and TME-related polygenic risk score (PRS). Secondly, according to the TP-related and TME-related PRSs, we proposed the TP-TME risk classification which was validated in two external data sets from The Cancer Genome Atlas Program and International Cancer Genome Consortium database. We also performed functional enrichment and drug repositioning analysis to reveal the potential biological heterogeneity among different subtypes. Results: The three TP-TME risk subtypes of HCC had significantly different prognosis and biological characteristics. The TP-TME low risk subtype had the best prognosis and was characterized by well-differentiated, the TP-TME high risk subtype had the worst prognosis and was characterized by aberrant activation of TGFβ and WNT pathways, and the TP-TME high risk subtype had the moderate prognosis and was characterized by exhibited activated MYC targets and proliferation-related gene sets. We also found that these three TP-TME risk subtypes may respond differently to immunotherapy (e.g., immune checkpoint inhibitors and chimeric antigen receptor-modified T cells) or other drug therapies. Conclusions: By combining separately constructed TP-related PRS and TME-related PRS, we proposed and validated a novel molecular classification system, the TP-TME risk subtyping system, to divide patients with HCC into three subtypes with distinct biological characteristics and prognosis. These findings highlight the significant clinical implications of the TP-TME risk subtyping system and provide potential personalized immunotherapy strategies for patients with HCC.

OmicsNet 2.0: a web-based platform for multi-omics integration and network visual analytics.

Researchers are increasingly seeking to interpret molecular data within a multi-omics context to gain a more comprehensive picture of their study system. OmicsNet (www.omicsnet.ca) is a web-based tool developed to allow users to easily build, visualize, and analyze multi-omics networks to study rich relationships among lists of ‘omics features of interest. Three major improvements have been introduced in OmicsNet 2.0, which include: (i) enhanced network visual analytics with eleven 2D graph layout options and a novel 3D module layout; (ii) support for three new ‘omics types: single nucleotide polymorphism (SNP) list from genetic variation studies; taxon list from microbiome profiling studies, as well as liquid chromatography–mass spectrometry (LC–MS) peaks from untargeted metabolomics; and (iii) measures to improve research reproducibility by coupling R command history with the release of the companion OmicsNetR package, and generation of persistent links to share interactive network views. We performed a case study using the multi-omics data obtained from a recent large-scale investigation on inflammatory bowel disease (IBD) and demonstrated that OmicsNet was able to quickly create meaningful multi-omics context to facilitate hypothesis generation and mechanistic insights.