PARADIGM-SHIFT predicts the function of mutations in multiple cancers using pathway impact analysis

MPAC: a computational framework for inferring pathway activities from multi-omic data

Fully capturing cellular state requires examining genomic, epigenomic, transcriptomic, proteomic, and other assays for a biological sample and comprehensive computational modeling to reason with the complex and sometimes conflicting measurements. Modeling these so-called multi-omic data is especially beneficial in disease analysis, where observations across omic data types may reveal unexpected patient groupings and inform clinical outcomes and treatments. We present Multi-omic Pathway Analysis of Cells (MPAC), a computational framework that interprets multi-omic data through prior knowledge from biological pathways. MPAC uses network relationships encoded in pathways using a factor graph to infer consensus activity levels for proteins and associated pathway entities from multi-omic data, runs permutation testing to eliminate spurious activity predictions, and groups biological samples by pathway activities to prioritize proteins with potential clinical relevance. Using DNA copy number alteration and RNA-seq data from head and neck squamous cell carcinoma patients from The Cancer Genome Atlas as an example, we demonstrate that MPAC predicts a patient subgroup related to immune responses not identified by analysis with either input omic data type alone. Key proteins identified via this subgroup have pathway activities related to clinical outcome as well as immune cell compositions. Our MPAC R package, available at https://bioconductor.org/packages/MPAC, enables similar multi-omic analyses on new datasets.

Integrative Computational Framework, Dyscovr, Links Mutated Driver Genes to Expression Dysregulation Across 19 Cancer Types

Though somatic mutations play a critical role in driving cancer initiation and progression, the systems-level functional impacts of these mutations-particularly, how they alter expression across the genome and give rise to cancer hallmarks-are not yet well-understood, even for well-studied cancer driver genes. To address this, we designed an integrative machine learning model, Dyscovr, that leverages mutation, gene expression, copy number alteration (CNA), methylation, and clinical data to uncover putative relationships between nonsynonymous mutations in key cancer driver genes and transcriptional changes across the genome. We applied Dyscovr pan-cancer and within 19 individual cancer types, finding both broadly relevant and cancer type-specific links between driver genes and putative targets, including a subset we further identify as exhibiting negative genetic relationships. Our work newly implicates-and validates in cell lines-KBTBD2 and mutant PIK3CA as putative synthetic lethals in breast cancer, suggesting a novel combinatorial treatment approach.

Advances in computational methods for identifying cancer driver genes

Cancer driver genes (CDGs) are crucial in cancer prevention, diagnosis and treatment. This study employed computational methods for identifying CDGs, categorizing them into four groups. The major frameworks for each of these four categories were summarized. Additionally, we systematically gathered data from public databases and biological networks, and we elaborated on computational methods for identifying CDGs using the aforementioned databases. Further, we summarized the algorithms, mainly involving statistics and machine learning, used for identifying CDGs. Notably, the performances of nine typical identification methods for eight types of cancer were compared to analyze the applicability areas of these methods. Finally, we discussed the challenges and prospects associated with methods for identifying CDGs. The present study revealed that the network-based algorithms and machine learning-based methods demonstrated superior performance.

Unravelling cancer subtype-specific driver genes in single-cell transcriptomics data with CSDGI

Cancer is known as a heterogeneous disease. Cancer driver genes (CDGs) need to be inferred for understanding tumor heterogeneity in cancer. However, the existing computational methods have identified many common CDGs. A key challenge exploring cancer progression is to infer cancer subtype-specific driver genes (CSDGs), which provides guidane for the diagnosis, treatment and prognosis of cancer. The significant advancements in single-cell RNA-sequencing (scRNA-seq) technologies have opened up new possibilities for studying human cancers at the individual cell level. In this study, we develop a novel unsupervised method, CSDGI (Cancer Subtype-specific Driver Gene Inference), which applies Encoder-Decoder-Framework consisting of low-rank residual neural networks to inferring driver genes corresponding to potential cancer subtypes at the single-cell level. To infer CSDGs, we apply CSDGI to the tumor single-cell transcriptomics data. To filter the redundant genes before driver gene inference, we perform the differential expression genes (DEGs). The experimental results demonstrate CSDGI is effective to infer driver genes that are cancer subtype-specific. Functional and disease enrichment analysis shows these inferred CSDGs indicate the key biological processes and disease pathways. CSDGI is the first method to explore cancer driver genes at the cancer subtype level. We believe that it can be a useful method to understand the mechanisms of cell transformation driving tumours.

Integrative genetic and genomic networks identify microRNA associated with COPD and ILD

Chronic obstructive pulmonary disease (COPD) and interstitial lung disease (ILD) are clinically and molecularly heterogeneous diseases. We utilized clustering and integrative network analyses to elucidate roles for microRNAs (miRNAs) and miRNA isoforms (isomiRs) in COPD and ILD pathogenesis. Short RNA sequencing was performed on 351 lung tissue samples of COPD (n = 145), ILD (n = 144) and controls (n = 64). Five distinct subclusters of samples were identified including 1 COPD-predominant cluster and 2 ILD-predominant clusters which associated with different clinical measurements of disease severity. Utilizing 262 samples with gene expression and SNP microarrays, we built disease-specific genetic and expression networks to predict key miRNA regulators of gene expression. Members of miR-449/34 family, known to promote airway differentiation by repressing the Notch pathway, were among the top connected miRNAs in both COPD and ILD networks. Genes associated with miR-449/34 members in the disease networks were enriched among genes that increase in expression with airway differentiation at an air-liquid interface. A highly expressed isomiR containing a novel seed sequence was identified at the miR-34c-5p locus. 47% of the anticorrelated predicted targets for this isomiR were distinct from the canonical seed sequence for miR-34c-5p. Overexpression of the canonical miR-34c-5p and the miR-34c-5p isomiR with an alternative seed sequence down-regulated NOTCH1 and NOTCH4. However, only overexpression of the isomiR down-regulated genes involved in Ras signaling such as CRKL and GRB2. Overall, these findings elucidate molecular heterogeneity inherent across COPD and ILD patients and further suggest roles for miR-34c in regulating disease-associated gene-expression.

Interpretable deep learning for chromatin-informed inference of transcriptional programs driven by somatic alterations across cancers

Cancer is a disease of gene dysregulation, where cells acquire somatic and epigenetic alterations that drive aberrant cellular signaling. These alterations adversely impact transcriptional programs and cause profound changes in gene expression. Interpreting somatic alterations within context-specific transcriptional programs will facilitate personalized therapeutic decisions but is a monumental task. Toward this goal, we develop a partially interpretable neural network model called Chromatin-informed Inference of Transcriptional Regulators Using Self-attention mechanism (CITRUS). CITRUS models the impact of somatic alterations on transcription factors and downstream transcriptional programs. Our approach employs a self-attention mechanism to model the contextual impact of somatic alterations. Furthermore, CITRUS uses a layer of hidden nodes to explicitly represent the state of transcription factors (TFs) to learn the relationships between TFs and their target genes based on TF binding motifs in the open chromatin regions of tumor samples. We apply CITRUS to genomic, transcriptomic, and epigenomic data from 17 cancer types profiled by The Cancer Genome Atlas. CITRUS predicts patient-specific TF activities and reveals transcriptional program variations between and within tumor types. We show that CITRUS yields biological insights into delineating TFs associated with somatic alterations in individual tumors. Thus, CITRUS is a promising tool for precision oncology.

Mutational signatures representative transcriptomic perturbations in hepatocellular carcinoma

Hepatocellular carcinoma (HCC) is a primary malignancy with increasing incidence and poor prognosis. Heterogeneity originating from genomic instability is one of the critical reasons of poor outcomes. However, the studies of underlying mechanisms and pathways affected by mutations are still not intelligible. Currently, integrative molecular-level studies using multiomics approaches enable comprehensive analysis for cancers, which is pivotal for personalized therapy and mortality reduction. In this study, genomic and transcriptomic data of HCC are obtained from The Cancer Genome Atlas (TCGA) to investigate the affected coding and non-coding RNAs, as well as their regulatory network due to certain mutational signatures of HCC. Different types of RNAs have their specific enriched biological functions in mutational signature-specific HCCs, upregulated coding RNAs are predominantly associated with lipid metabolism-related pathways, and downregulated coding RNAs are enriched in axonogenesis for tumor microenvironment generation. Additionally, differentially expressed miRNAs are inclined to concentrate in cancer-related signaling pathways. Some of these RNAs also serve as prognostic factors that help predict the survival outcome of HCCs with certain mutational signatures. Furthermore, deregulation of competing endogenous RNA (ceRNA) regulatory network is identified, which suggests a potential therapy via interference of miRNA activity for mutational signature-specific HCC. This study proposes a projection approach to reduce therapeutic complexity from genomic mutations to transcriptomic alterations. Through this method, we identify genes and pathways critical for mutational signature-specific HCC and further discover a series of prognostic markers indicating patient survival outcome.

Identification of a Five-Gene Panel to Assess Prognosis for Gastric Cancer

Methods

Two datasets were used as training and validation cohorts to establish the predictive model. We used three types of screening criteria: background analysis, pathway analysis, and functional analysis provided by the cBioportal website. Fisher's exact test and multivariable logistic regression were performed to screen out related genes. Furthermore, we performed receiver operating characteristic (ROC) and Kaplan–Meier curve analyses to evaluate the correlation between the selected genes and overall survival.

Result

We screened five genes (KNL1, NRXN1, C6, CCDC169-SOHLH2, and TTN) that were highly related to recurrence of GC. The area under the receiver operating characteristic (ROC) curve was 0.813, which was much higher than that of the baseline model (AUC = 0.699). This result suggested that the mutation of five selected genes had a significant effect on the prediction of recurrence compared with other factors (age, stages, history, etc.). Furthermore, the Kaplan-Meier estimator also revealed that the mutation of five genes positively correlated with patient survival.

Conclusions

The patients who have mutations in these five genes may experience longer survival than those who do not have mutations. This five-gene panel will likely be a practical tool for prognostic evaluation and will provide another possible way for clinicians to determine therapy.

Identification of driver genes based on gene mutational effects and network centrality

BACKGROUND

As one of the deadliest diseases in the world, cancer is driven by a few somatic mutations that disrupt the normal growth of cells, and leads to abnormal proliferation and tumor development. The vast majority of somatic mutations did not affect the occurrence and development of cancer; thus, identifying the mutations responsible for tumor occurrence and development is one of the main targets of current cancer treatments.

RESULTS

To effectively identify driver genes, we adopted a semi-local centrality measure and gene mutation effect function to assess the effect of gene mutations on changes in gene expression patterns. Firstly, we calculated the mutation score for each gene. Secondly, we identified differentially expressed genes (DEGs) in the cohort by comparing the expression profiles of tumor samples and normal samples, and then constructed a local network for each mutation gene using DEGs and mutant genes according to the protein-protein interaction network. Finally, we calculated the score of each mutant gene according to the objective function. The top-ranking mutant genes were selected as driver genes. We name the proposed method as mutations effect and network centrality.

CONCLUSIONS

Four types of cancer data in The Cancer Genome Atlas were tested. The experimental data proved that our method was superior to the existing network-centric method, as it was able to quickly and easily identify driver genes and rare driver factors.

Network control principles for identifying personalized driver genes in cancer

To understand tumor heterogeneity in cancer, personalized driver genes (PDGs) need to be identified for unraveling the genotype-phenotype associations corresponding to particular patients. However, most of the existing driver-focus methods mainly pay attention on the cohort information rather than on individual information. Recent developing computational approaches based on network control principles are opening a new way to discover driver genes in cancer, particularly at an individual level. To provide comprehensive perspectives of network control methods on this timely topic, we first considered the cancer progression as a network control problem, in which the expected PDGs are altered genes by oncogene activation signals that can change the individual molecular network from one health state to the other disease state. Then, we reviewed the network reconstruction methods on single samples and introduced novel network control methods on single-sample networks to identify PDGs in cancer. Particularly, we gave a performance assessment of the network structure control-based PDGs identification methods on multiple cancer datasets from TCGA, for which the data and evaluation package also are publicly available. Finally, we discussed future directions for the application of network control methods to identify PDGs in cancer and diverse biological processes.

A Hierarchical Graph Convolution Network for Representation Learning of Gene Expression Data

The curse of dimensionality, which is caused by high-dimensionality and low-sample-size, is a major challenge in gene expression data analysis. However, the real situation is even worse: labelling data is laborious and time-consuming, so only a small part of the limited samples will be labelled. Having such few labelled samples further increases the difficulty of training deep learning models. Interpretability is an important requirement in biomedicine. Many existing deep learning methods are trying to provide interpretability, but rarely apply to gene expression data. Recent semi-supervised graph convolution network methods try to address these problems by smoothing the label information over a graph. However, to the best of our knowledge, these methods only utilize graphs in either the feature space or sample space, which restrict their performance. We propose a transductive semi-supervised representation learning method called a hierarchical graph convolution network (HiGCN) to aggregate the information of gene expression data in both feature and sample spaces. HiGCN first utilizes external knowledge to construct a feature graph and a similarity kernel to construct a sample graph. Then, two spatial-based GCNs are used to aggregate information on these graphs. To validate the model's performance, synthetic and real datasets are provided to lend empirical support. Compared with two recent models and three traditional models, HiGCN learns better representations of gene expression data, and these representations improve the performance of downstream tasks, especially when the model is trained on a few labelled samples. Important features can be extracted from our model to provide reliable interpretability.

Identification of Breast Cancer Subtype-Specific Biomarkers by Integrating Copy Number Alterations and Gene Expression Profiles

Background and Objectives: Breast cancer is a heterogeneous disease categorized into four subtypes. Previous studies have shown that copy number alterations of several genes are implicated with the development and progression of many cancers. This study evaluates the effects of DNA copy number alterations on gene expression levels in different breast cancer subtypes. Materials and Methods: We performed a computational analysis integrating copy number alterations and gene expression profiles in 1024 breast cancer samples grouped into four molecular subtypes: luminal A, luminal B, HER2, and basal. Results: Our analyses identified several genes correlated in all subtypes such as KIAA1967 and MCPH1. In addition, several subtype-specific genes that showed a significant correlation between copy number and gene expression profiles were detected: SMARCB1, AZIN1, MTDH in luminal A, PPP2R5E, APEX1, GCN5 in luminal B, TNFAIP1, PCYT2, DIABLO in HER2, and FAM175B, SENP5, SCAF1 in basal subtype. Conclusions: This study showed that computational analyses integrating copy number and gene expression can contribute to unveil the molecular mechanisms of cancer and identify new subtype-specific biomarkers.

Pan-cancer analysis of somatic mutations in miRNA genes

Background

miRNAs are considered important players in oncogenesis, serving either as oncomiRs or suppressormiRs. Although the accumulation of somatic alterations is an intrinsic aspect of cancer development and many important cancer-driving mutations have been identified in protein-coding genes, the area of functional somatic mutations in miRNA genes is heavily understudied.

Methods

Here, based on the analysis of large genomic datasets, mostly the whole-exome sequencing of over 10,000 cancer/normal sample pairs deposited within the TCGA repository, we undertook an analysis of somatic mutations in miRNA genes.

Findings

We identified and characterized over 10,000 somatic mutations and showed that some of the miRNA genes are overmutated in Pan-Cancer and/or specific cancers. Nonrandom occurrence of the identified mutations was confirmed by a strong association of overmutated miRNA genes with KEGG pathways, most of which were related to specific cancer types or cancer-related processes. Additionally, we showed that mutations in some of the overmutated genes correlate with miRNA expression, cancer staging, and patient survival.

Interpretation

Our study is the first comprehensive Pan-Cancer study of cancer somatic mutations in miRNA genes. It may help to understand the consequences of mutations in miRNA genes and the identification of miRNA functional mutations. The results may also be the first step (form the basis and provide the resources) in the development of computational and/or statistical approaches/tools dedicated to the identification of cancer-driver miRNA genes.

Funding

This work was supported by research grants from the Polish National Science Centre 2016/22/A/NZ2/00184 and 2015/17/N/NZ3/03629.

CNet: a multi-omics approach to detecting clinically associated, combinatory genomic signatures

MOTIVATION

Genome-wide multi-omics profiling of complex diseases provides valuable resources and opportunities to discover associations between various measures of genes and diseases. Currently, a pressing challenge is how to effectively detect functional genes associated with or causing phenotypic outcomes. We developed CNet to identify groups of genomic signatures whose combinatory effect is significantly associated with clinical and phenotypical outcomes.

RESULTS

CNet builds on a generalized sequential feedforward method, augmented by a down-sampling bootstrap strategy to reduce random hitchhiking signatures. It further applies a dynamic trimming procedure to remove relatively less informative signatures at every step. CNet can manage heterogeneous genomic signature profiles simultaneously and select the best signature to represent a specific gene. To deal with various forms of clinical and phenotypical measurements, we introduced four models to deal with continuous, categorical and censored data. We tested CNet using drug-response data, multidimensional cancer genomics data and genome-wide association study data for multiple traits. Our results demonstrated that in various scenarios, CNet could effectively identify signatures that are associated with the outcomes. In addition, we applied CNet to identify likely disease-causing chains involving somatic mutations, pathway activities and patient outcomes. With appropriate setting, CNet can be applied in many biological conditions.

AVAILABILITY AND IMPLEMENTATION

CNet can be downloaded at https://github.com/bsml320/CNet.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

PRODIGY: personalized prioritization of driver genes

MOTIVATION

Evolution of cancer is driven by few somatic mutations that disrupt cellular processes, causing abnormal proliferation and tumor development, whereas most somatic mutations have no impact on progression. Distinguishing those mutated genes that drive tumorigenesis in a patient is a primary goal in cancer therapy: Knowledge of these genes and the pathways on which they operate can illuminate disease mechanisms and indicate potential therapies and drug targets. Current research focuses mainly on cohort-level driver gene identification but patient-specific driver gene identification remains a challenge.

METHODS

We developed a new algorithm for patient-specific ranking of driver genes. The algorithm, called PRODIGY, analyzes the expression and mutation profiles of the patient along with data on known pathways and protein-protein interactions. Prodigy quantifies the impact of each mutated gene on every deregulated pathway using the prize-collecting Steiner tree model. Mutated genes are ranked by their aggregated impact on all deregulated pathways.

RESULTS

In testing on five TCGA cancer cohorts spanning >2500 patients and comparison to validated driver genes, Prodigy outperformed extant methods and ranking based on network centrality measures. Our results pinpoint the pleiotropic effect of driver genes and show that Prodigy is capable of identifying even very rare drivers. Hence, Prodigy takes a step further toward personalized medicine and treatment.

AVAILABILITY AND IMPLEMENTATION

The Prodigy R package is available at: https://github.com/Shamir-Lab/PRODIGY.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Dissecting the sources of gene expression variation in a pan-cancer analysis identifies novel regulatory mutations

Although the catalog of cancer-associated mutations in protein-coding regions is nearly complete for all major cancer types, an assessment of regulatory changes in cancer genomes and their clinical significance remain largely preliminary. Adopting bottom-up approach, we quantify the effects of different sources of gene expression variation in a cohort of 3899 samples from 10 cancer types. We find that copy number alterations, epigenetic changes, transcription factors and microRNAs collectively explain, on average, only 31–38% and 18–26% expression variation for cancer-associated and other genes, respectively, and that among these factors copy number alteration has the highest effect. We show that the genes with systematic, large expression variation that could not be attributed to these factors are enriched for pathways related to cancer hallmarks. Integrating whole genome sequencing data and focusing on genes with systematic expression variation we identify novel, recurrent regulatory mutations affecting known cancer genes such as NKX2-1 and GRIN2D in multiple cancer types. Nonetheless, at a genome-wide scale proportions of gene expression variation attributed to recurrent point mutations appear to be modest so far, especially when compared to that attributed to copy number changes – a pattern different from that observed for other complex diseases and traits. We suspect that, owing to plasticity and redundancy in biological pathways, regulatory alterations show complex combinatorial patterns, modulating gene expression in cancer genomes at a finer scale.

Multi-omic and multi-view clustering algorithms: review and cancer benchmark

Recent high throughput experimental methods have been used to collect large biomedical omics datasets. Clustering of single omic datasets has proven invaluable for biological and medical research. The decreasing cost and development of additional high throughput methods now enable measurement of multi-omic data. Clustering multi-omic data has the potential to reveal further systems-level insights, but raises computational and biological challenges. Here, we review algorithms for multi-omics clustering, and discuss key issues in applying these algorithms. Our review covers methods developed specifically for omic data as well as generic multi-view methods developed in the machine learning community for joint clustering of multiple data types. In addition, using cancer data from TCGA, we perform an extensive benchmark spanning ten different cancer types, providing the first systematic comparison of leading multi-omics and multi-view clustering algorithms. The results highlight key issues regarding the use of single- versus multi-omics, the choice of clustering strategy, the power of generic multi-view methods and the use of approximated p-values for gauging solution quality. Due to the growing use of multi-omics data, we expect these issues to be important for future progress in the field.

A Novel Method for Identifying the Potential Cancer Driver Genes Based on Molecular Data Integration

The identification of the cancer driver genes is essential for personalized therapy. The mutation frequency of most driver genes is in the middle (2-20%) or even lower range, which makes it difficult to find the driver genes with low-frequency mutations. Other forms of genomic aberrations, such as copy number variations (CNVs) and epigenetic changes, may also reflect cancer progression. In this work, a method for identifying the potential cancer driver genes (iPDG) based on molecular data integration is proposed. DNA copy number variation, somatic mutation, and gene expression data of matched cancer samples are integrated. In combination with the method of iKEEG, the "key genes" of cancer are identified, and the change in their expression levels is used for auxiliary evaluation of whether the mutated genes are potential drivers. For a mutated gene, the concept of mutational effect is defined, which takes into account the effects of copy number variation, mutation gene itself, and its neighbor genes. The method mainly includes two steps: the first step is data preprocessing. First, DNA copy number variation and somatic mutation data are integrated. Then, the integrated data are mapped to a given interaction network, and the diffusion kernel is used to form the mutation effect matrix. The second step is to obtain the key genes by using the iKGGE method, and construct the connection matrix by means of the gene expression data of the key genes and mutation impact matrix of the mutated genes. Experiments on TCGA breast cancer and Glioblastoma multiforme datasets demonstrate that iPDG is effective not only to identify the known cancer driver genes but also to discover the rare potential driver genes. When measured by functional enrichment analysis, we find that these genes are clearly associated with these two types of cancers.

Associating expression and genomic data using co-occurrence measures

Abstract

Recent technological evolutions have led to an exponential increase in data in all the omics fields. It is expected that integration of these different data sources, will drastically enhance our knowledge of the biological mechanisms behind genomic diseases such as cancer. However, the integration of different omics data still remains a challenge. In this work we propose an intuitive workflow for the integrative analysis of expression, mutation and copy number data taken from the METABRIC study on breast cancer. First, we present evidence that the expression profile of many important breast cancer genes consists of two modes or ‘regimes’, which contain important clinical information. Then, we show how the co-occurrence of these expression regimes can be used as an association measure between genes and validate our findings on the TCGA-BRCA study. Finally, we demonstrate how these co-occurrence measures can also be applied to link expression regimes to genomic aberrations, providing a more complete, integrative view on breast cancer. As a case study, an integrative analysis of the identified MLPH-FOXA1 association is performed, illustrating that the obtained expression associations are intimately linked to the underlying genomic changes.

Reviewers

This article was reviewed by Dirk Walther, Francisco Garcia and Isabel Nepomuceno.

Electronic supplementary material

The online version of this article (10.1186/s13062-019-0240-2) contains supplementary material, which is available to authorized users.

CaDrA: A Computational Framework for Performing Candidate Driver Analyses Using Genomic Features

The identification of genetic alteration combinations as drivers of a given phenotypic outcome, such as drug sensitivity, gene or protein expression, and pathway activity, is a challenging task that is essential to gaining new biological insights and to discovering therapeutic targets. Existing methods designed to predict complementary drivers of such outcomes lack analytical flexibility, including the support for joint analyses of multiple genomic alteration types, such as somatic mutations and copy number alterations, multiple scoring functions, and rigorous significance and reproducibility testing procedures. To address these limitations, we developed Candidate Driver Analysis or CaDrA, an integrative framework that implements a step-wise heuristic search approach to identify functionally relevant subsets of genomic features that, together, are maximally associated with a specific outcome of interest. We show CaDrA's overall high sensitivity and specificity for typically sized multi-omic datasets using simulated data, and demonstrate CaDrA's ability to identify known mutations linked with sensitivity of cancer cells to drug treatment using data from the Cancer Cell Line Encyclopedia (CCLE). We further apply CaDrA to identify novel regulators of oncogenic activity mediated by Hippo signaling pathway effectors YAP and TAZ in primary breast cancer tumors using data from The Cancer Genome Atlas (TCGA), which we functionally validate in vitro. Finally, we use pan-cancer TCGA protein expression data to show the high reproducibility of CaDrA's search procedure. Collectively, this work demonstrates the utility of our framework for supporting the fast querying of large, publicly available multi-omics datasets, including but not limited to TCGA and CCLE, for potential drivers of a given target profile of interest.

A multi-omics approach for identifying important pathways and genes in human cancer

Background

Cancer develops when pathways controlling cell survival, cell fate or genome maintenance are disrupted by the somatic alteration of key driver genes. Understanding how pathway disruption is driven by somatic alterations is thus essential for an accurate characterization of cancer biology and identification of therapeutic targets. Unfortunately, current cancer pathway analysis methods fail to fully model the relationship between somatic alterations and pathway activity.

Results

To address these limitations, we developed a multi-omics method for identifying biologically important pathways and genes in human cancer. Our approach combines single-sample pathway analysis with multi-stage, lasso-penalized regression to find pathways whose gene expression can be explained largely in terms of gene-level somatic alterations in the tumor. Importantly, this method can analyze case-only data sets, does not require information regarding pathway topology and supports personalized pathway analysis using just somatic alteration data for a limited number of cancer-associated genes. The practical effectiveness of this technique is illustrated through an analysis of data from The Cancer Genome Atlas using gene sets from the Molecular Signatures Database.

Conclusions

Novel insights into the pathophysiology of human cancer can be obtained from statistical models that predict expression-based pathway activity in terms of non-silent somatic mutations and copy number variation. These models enable the identification of biologically important pathways and genes and support personalized pathway analysis in cases where gene expression data is unavailable.

Electronic supplementary material

The online version of this article (10.1186/s12859-018-2476-8) contains supplementary material, which is available to authorized users.

Personalized Integrated Network Modeling of the Cancer Proteome Atlas

Personalized (patient-specific) approaches have recently emerged with a precision medicine paradigm that acknowledges the fact that molecular pathway structures and activity might be considerably different within and across tumors. The functional cancer genome and proteome provide rich sources of information to identify patient-specific variations in signaling pathways and activities within and across tumors; however, current analytic methods lack the ability to exploit the diverse and multi-layered architecture of these complex biological networks. We assessed pan-cancer pathway activities for >7700 patients across 32 tumor types from The Cancer Proteome Atlas by developing a personalized cancer-specific integrated network estimation (PRECISE) model. PRECISE is a general Bayesian framework for integrating existing interaction databases, data-driven de novo causal structures, and upstream molecular profiling data to estimate cancer-specific integrated networks, infer patient-specific networks and elicit interpretable pathway-level signatures. PRECISE-based pathway signatures, can delineate pan-cancer commonalities and differences in proteomic network biology within and across tumors, demonstrates robust tumor stratification that is both biologically and clinically informative and superior prognostic power compared to existing approaches. Towards establishing the translational relevance of the functional proteome in research and clinical settings, we provide an online, publicly available, comprehensive database and visualization repository of our findings ( https://mjha.shinyapps.io/PRECISE/ ).

The Emerging Potential for Network Analysis to Inform Precision Cancer Medicine

Precision cancer medicine promises to tailor clinical decisions to patients using genomic information. Indeed, successes of drugs targeting genetic alterations in tumors, such as imatinib that targets BCR-ABL in chronic myelogenous leukemia, have demonstrated the power of this approach. However, biological systems are complex, and patients may differ not only by the specific genetic alterations in their tumor, but also by more subtle interactions among such alterations. Systems biology and more specifically, network analysis, provides a framework for advancing precision medicine beyond clinical actionability of individual mutations. Here we discuss applications of network analysis to study tumor biology, early methods for N-of-1 tumor genome analysis, and the path for such tools to the clinic.

FusionPathway: Prediction of pathways and therapeutic targets associated with gene fusions in cancer

Numerous gene fusions have been uncovered across multiple cancer types. Although the ability to target several of these fusions has led to the development of some successful anti-cancer drugs, most of them are not druggable. Understanding the molecular pathways of a fusion is important in determining its function in oncogenesis and in developing therapeutic strategies for patients harboring the fusion. However, the molecular pathways have been elucidated for only a few fusions, in part because of the labor-intensive nature of the required functional assays. Therefore, we developed a domain-based network approach to infer the pathways of a fusion. Molecular interactions of a fusion are first predicted by using its protein domain composition, and its associated pathways are then inferred from these molecular interactions. We demonstrated the capabilities of this approach by primarily applying it to the well-studied BCR-ABL1 fusion. The approach was also applied to two undruggable fusions in sarcoma, EWS-FL1 and FUS-DDIT3. We successfully identified known genes and pathways associated with these fusions and satisfactorily validated these predictions using several benchmark sets. The predictions of EWS-FL1 and FUS-DDIT3 also correlate with results of high-throughput drug screening. To our best knowledge, this is the first approach for inferring pathways of fusions.

We present a computational framework, FusionPathway, to infer the oncogenesis pathways of a fusion and help develop therapeutic strategies in these pathways for patients harboring the fusion. In this work, we successfully validated the capabilities of this approach through its application to the well-studied BCR-ABL1 fusion and two undruggable fusions in sarcoma, EWS-FL1 and FUS-DDIT3. Especially, the predictions of EWS-FL1 and FUS-DDIT3 correlate well with results of high-throughput drug screening in sarcoma cells. Therefore, FusionPathway can be an effective method to infer pathways and potential therapeutic targets that are associated with those undruggable fusions. Our results of BCR-ABL1 also suggest that FusionPathway may be able to help elucidate pathway-dependent mechanisms of resistances to those kinase fusion-targeting therapies and develop strategies to overcome the resistances. In addition, the developed R package of FusionPathways (https://github.com/perwu/FusionPathway/) can help people easily apply our approach to study other important fusions in cancer.

Integrative landscape of dysregulated signaling pathways of clinically distinct pancreatic cancer subtypes

Despite modern therapeutic advances, the survival prospects of pancreatic cancer patients have remained poor. Besides being highly metastatic, pancreatic cancer is challenging to treat because it is caused by a heterogeneous array of somatic mutations that impact a variety of signaling pathways and cellular regulatory systems. Here we use publicly available transcriptomic, copy number alteration and mutation profiling datasets from pancreatic cancer patients together with data on disease outcomes to show that the three major pancreatic cancer subtypes each display distinctive aberrations in cell signaling and metabolic pathways. Importantly, patients afflicted with these different pancreatic cancer subtypes also exhibit distinctive survival profiles. Within these patients, we find that dysregulation of the phosphoinositide 3-kinase and mitogen-activated protein kinase pathways, and p53 mediated disruptions of cell cycle processes are apparently drivers of disease. Further, we identify for the first time the molecular perturbations of signalling networks that are likely the primary causes of oncogenesis in each of the three pancreatic cancer subtypes.

Machine Learning Detects Pan-cancer Ras Pathway Activation in The Cancer Genome Atlas

Precision oncology uses genomic evidence to match patients with treatment but often fails to identify all patients who may respond. The transcriptome of these "hidden responders" may reveal responsive molecular states. We describe and evaluate a machine-learning approach to classify aberrant pathway activity in tumors, which may aid in hidden responder identification. The algorithm integrates RNA-seq, copy number, and mutations from 33 different cancer types across The Cancer Genome Atlas (TCGA) PanCanAtlas project to predict aberrant molecular states in tumors. Applied to the Ras pathway, the method detects Ras activation across cancer types and identifies phenocopying variants. The model, trained on human tumors, can predict response to MEK inhibitors in wild-type Ras cell lines. We also present data that suggest that multiple hits in the Ras pathway confer increased Ras activity. The transcriptome is underused in precision oncology and, combined with machine learning, can aid in the identification of hidden responders.

Practical aspects of NGS-based pathways analysis for personalized cancer science and medicine

Nowadays, the personalized approach to health care and cancer care in particular is becoming more and more popular and is taking an important place in the translational medicine paradigm. In some cases, detection of the patient-specific individual mutations that point to a targeted therapy has already become a routine practice for clinical oncologists. Wider panels of genetic markers are also on the market which cover a greater number of possible oncogenes including those with lower reliability of resulting medical conclusions. In light of the large availability of high-throughput technologies, it is very tempting to use complete patient-specific New Generation Sequencing (NGS) or other "omics" data for cancer treatment guidance. However, there are still no gold standard methods and protocols to evaluate them. Here we will discuss the clinical utility of each of the data types and describe a systems biology approach adapted for single patient measurements. We will try to summarize the current state of the field focusing on the clinically relevant case-studies and practical aspects of data processing.

Identification of cancer genes that are independent of dominant proliferation and lineage programs

Large, multidimensional “landscaping” projects have provided datasets that can be mined to identify potential targets for subgroups of tumors. Here, we analyzed genomic and transcriptomic data from human breast tumors to identify genes whose expression is enriched in tumors harboring specific genetic alterations. However, this analysis revealed that two other factors, proliferation rate and tumor lineage, are more dominant factors in shaping tumor transcriptional programs than genetic alterations. This discovery shifted our attention to identifying genes that are independent of the dominant proliferation and lineage programs. A small subset of these genes represents candidate targets for combination cancer therapies because they are druggable, maintained after treatment with chemotherapy, essential for cell line survival, and elevated in drug-resistant stem-like cancer cells.

Large, multidimensional cancer datasets provide a resource that can be mined to identify candidate therapeutic targets for specific subgroups of tumors. Here, we analyzed human breast cancer data to identify transcriptional programs associated with tumors bearing specific genetic driver alterations. Using an unbiased approach, we identified thousands of genes whose expression was enriched in tumors with specific genetic alterations. However, expression of the vast majority of these genes was not enriched if associations were analyzed within individual breast tumor molecular subtypes, across multiple tumor types, or after gene expression was normalized to account for differences in proliferation or tumor lineage. Together with linear modeling results, these findings suggest that most transcriptional programs associated with specific genetic alterations in oncogenes and tumor suppressors are highly context-dependent and are predominantly linked to differences in proliferation programs between distinct breast cancer subtypes. We demonstrate that such proliferation-dependent gene expression dominates tumor transcriptional programs relative to matched normal tissues. However, we also identified a relatively small group of cancer-associated genes that are both proliferation- and lineage-independent. A subset of these genes are attractive candidate targets for combination therapy because they are essential in breast cancer cell lines, druggable, enriched in stem-like breast cancer cells, and resistant to chemotherapy-induced down-regulation.

Impacts of somatic mutations on gene expression: an association perspective

Assessing the functional impacts of somatic mutations in cancer genomes is critical for both identifying driver mutations and developing molecular targeted therapies. Currently, it remains a fundamental challenge to distinguish the patterns through which mutations execute their biological effects and to infer biological mechanisms underlying these patterns. To this end, we systematically studied the association between somatic mutations in protein-coding regions and expression profiles, which represents an indirect measurement of impacts. We defined mutation features (mutation type, cluster and status) and built linear regression models to assess mutation associations with mRNA expression and protein expression. Our results presented a comprehensive landscape of the associations between mutation features and expression profile in multiple cancer types, including 62 genes showing mutation type associated expression changes, 21 genes showing mutation cluster associations and 51 genes showing mutation status associations. We revealed four characteristics of the patterns that mutations impact on expression. First, we showed that mutation type (truncation versus amino acid-altering mutations) was the most important determinant of expression levels. Second, we detected mutation clusters in well-studied oncogenes that were associated with gene expression. Third, we found both similarities and differences in association patterns existed within and across cancer types. Fourth, although many of the observed associations stay stable at both mRNA and protein expression levels, there are also novel associations uniquely observed at the protein level, which warrant future investigation. Taken together, our findings provided implications for cancer driver gene prioritization and insights into the functional consequences of somatic mutations.

DGPathinter: a novel model for identifying driver genes via knowledge-driven matrix factorization with prior knowledge from interactome and pathways

Cataloging mutated driver genes that confer a selective growth advantage for tumor cells from sporadic passenger mutations is a critical problem in cancer genomic research. Previous studies have reported that some driver genes are not highly frequently mutated and cannot be tested as statistically significant, which complicates the identification of driver genes. To address this issue, some existing approaches incorporate prior knowledge from an interactome to detect driver genes which may be dysregulated by interaction network context. However, altered operations of many pathways in cancer progression have been frequently observed, and prior knowledge from pathways is not exploited in the driver gene identification task. In this paper, we introduce a driver gene prioritization method called driver gene identification through pathway and interactome information (DGPathinter), which is based on knowledge-based matrix factorization model with prior knowledge from both interactome and pathways incorporated. When DGPathinter is applied on somatic mutation datasets of three types of cancers and evaluated by known driver genes, the prioritizing performances of DGPathinter are better than the existing interactome driven methods. The top ranked genes detected by DGPathinter are also significantly enriched for known driver genes. Moreover, most of the top ranked scored pathways given by DGPathinter are also cancer progression-associated pathways. These results suggest that DGPathinter is a useful tool to identify potential driver genes.

Functional variomics and network perturbation: connecting genotype to phenotype in cancer

Proteins interact with other macromolecules in complex cellular networks for signal transduction and biological function. In cancer, genetic aberrations have been traditionally thought to disrupt the entire gene function. It has been increasingly appreciated that each mutation of a gene could have a subtle but unique effect on protein function or network rewiring, contributing to diverse phenotypic consequences across cancer patient populations. In this Review, we discuss the current understanding of cancer genetic variants, including the broad spectrum of mutation classes and the wide range of mechanistic effects on gene function in the context of signalling networks. We highlight recent advances in computational and experimental strategies to study the diverse functional and phenotypic consequences of mutations at the base-pair resolution. Such information is crucial to understanding the complex pleiotropic effect of cancer genes and provides a possible link between genotype and phenotype in cancer.

Distribution bias analysis of germline and somatic single-nucleotide variations that impact protein functional site and neighboring amino acids

Single nucleotide variations (SNVs) can result in loss or gain of protein functional sites. We analyzed the effects of SNVs on enzyme active sites, ligand binding sites, and various types of post translational modification (PTM) sites. We found that, for most types of protein functional sites, the SNV pattern differs between germline and somatic mutations as well as between synonymous and non-synonymous mutations. From a total of 51,138 protein functional site affecting SNVs (pfsSNVs), a pan-cancer analysis revealed 142 somatic pfsSNVs in five or more cancer types. By leveraging patient information for somatic pfsSNVs, we identified 17 loss of functional site SNVs and 60 gain of functional site SNVs which are significantly enriched in patients with specific cancer types. Of the key pfsSNVs identified in our analysis above, we highlight 132 key pfsSNVs within 17 genes that are found in well-established cancer associated gene lists. For illustrating how key pfsSNVs can be prioritized further, we provide a use case where we performed survival analysis showing that a loss of phosphorylation site pfsSNV at position 105 in MEF2A is significantly associated with decreased pancreatic cancer patient survival rate. These 132 pfsSNVs can be used in developing genetic testing pipelines.

Discovering potential driver genes through an integrated model of somatic mutation profiles and gene functional information

An integrated approach to identify driver genes based on information of somatic mutations, the interaction network and Gene Ontology similarity.

iCAGES: integrated CAncer GEnome Score for comprehensively prioritizing driver genes in personal cancer genomes

Cancer results from the acquisition of somatic driver mutations. Several computational tools can predict driver genes from population-scale genomic data, but tools for analyzing personal cancer genomes are underdeveloped. Here we developed iCAGES, a novel statistical framework that infers driver variants by integrating contributions from coding, non-coding, and structural variants, identifies driver genes by combining genomic information and prior biological knowledge, then generates prioritized drug treatment. Analysis on The Cancer Genome Atlas (TCGA) data showed that iCAGES predicts whether patients respond to drug treatment (P = 0.006 by Fisher's exact test) and long-term survival (P = 0.003 from Cox regression). iCAGES is available at http://icages.wglab.org .

SSA-ME Detection of cancer driver genes using mutual exclusivity by small subnetwork analysis

Because of its clonal evolution a tumor rarely contains multiple genomic alterations in the same pathway as disrupting the pathway by one gene often is sufficient to confer the complete fitness advantage. As a result, many cancer driver genes display mutual exclusivity across tumors. However, searching for mutually exclusive gene sets requires analyzing all possible combinations of genes, leading to a problem which is typically too computationally complex to be solved without a stringent a priori filtering, restricting the mutations included in the analysis. To overcome this problem, we present SSA-ME, a network-based method to detect cancer driver genes based on independently scoring small subnetworks for mutual exclusivity using a reinforced learning approach. Because of the algorithmic efficiency, no stringent upfront filtering is required. Analysis of TCGA cancer datasets illustrates the added value of SSA-ME: well-known recurrently mutated but also rarely mutated drivers are prioritized. We show that using mutual exclusivity to detect cancer driver genes is complementary to state-of-the-art approaches. This framework, in which a large number of small subnetworks are being analyzed in order to solve a computationally complex problem (SSA), can be generically applied to any problem in which local neighborhoods in a network hold useful information.

TNF-α modulates genome-wide redistribution of ΔNp63α/TAp73 and NF-κB cREL interactive binding on TP53 and AP-1 motifs to promote an oncogenic gene program in squamous cancer

The Cancer Genome Atlas (TCGA) network study of 12 cancer types (PanCancer 12) revealed frequent mutation of TP53, and amplification and expression of related TP63 isoform ΔNp63 in squamous cancers. Further, aberrant expression of inflammatory genes and TP53/p63/p73 targets were detected in the PanCancer 12 project, reminiscent of gene programs comodulated by cREL/ΔNp63/TAp73 transcription factors we uncovered in head and neck squamous cell carcinomas (HNSCCs). However, how inflammatory gene signatures and cREL/p63/p73 targets are comodulated genome wide is unclear. Here, we examined how the inflammatory factor tumor necrosis factor-α (TNF-α) broadly modulates redistribution of cREL with ΔNp63α/TAp73 complexes and signatures genome wide in the HNSCC model UM-SCC46 using chromatin immunoprecipitation sequencing (ChIP-seq). TNF-α enhanced genome-wide co-occupancy of cREL with ΔNp63α on TP53/p63 sites, while unexpectedly promoting redistribution of TAp73 from TP53 to activator protein-1 (AP-1) sites. cREL, ΔNp63α and TAp73 binding and oligomerization on NF-κB-, TP53- or AP-1-specific sequences were independently validated by ChIP-qPCR (quantitative PCR), oligonucleotide-binding assays and analytical ultracentrifugation. Function of the binding activity was confirmed using TP53-, AP-1- and NF-κB-specific REs or p21, SERPINE1 and IL-6 promoter luciferase reporter activities. Concurrently, TNF-α regulated a broad gene network with cobinding activities for cREL, ΔNp63α and TAp73 observed upon array profiling and reverse transcription-PCR. Overlapping target gene signatures were observed in squamous cancer subsets and in inflamed skin of transgenic mice overexpressing ΔNp63α. Furthermore, multiple target genes identified in this study were linked to TP63 and TP73 activity and increased gene expression in large squamous cancer samples from PanCancer 12 TCGA by CircleMap. PARADIGM inferred pathway analysis revealed the network connection of TP63 and NF-κB complexes through an AP-1 hub, further supporting our findings. Thus, inflammatory cytokine TNF-α mediates genome-wide redistribution of the cREL/p63/p73, and AP-1 interactome, to diminish TAp73 tumor suppressor function and reciprocally activate NF-κB and AP-1 gene programs implicated in malignancy.

Harnessing Big Data for Systems Pharmacology

Systems pharmacology aims to holistically understand mechanisms of drug actions to support drug discovery and clinical practice. Systems pharmacology modeling (SPM) is data driven. It integrates an exponentially growing amount of data at multiple scales (genetic, molecular, cellular, organismal, and environmental). The goal of SPM is to develop mechanistic or predictive multiscale models that are interpretable and actionable. The current explosions in genomics and other omics data, as well as the tremendous advances in big data technologies, have already enabled biologists to generate novel hypotheses and gain new knowledge through computational models of genome-wide, heterogeneous, and dynamic data sets. More work is needed to interpret and predict a drug response phenotype, which is dependent on many known and unknown factors. To gain a comprehensive understanding of drug actions, SPM requires close collaborations between domain experts from diverse fields and integration of heterogeneous models from biophysics, mathematics, statistics, machine learning, and semantic webs. This creates challenges in model management, model integration, model translation, and knowledge integration. In this review, we discuss several emergent issues in SPM and potential solutions using big data technology and analytics. The concurrent development of high-throughput techniques, cloud computing, data science, and the semantic web will likely allow SPM to be findable, accessible, interoperable, reusable, reliable, interpretable, and actionable.

Non-linear Bayesian framework to determine the transcriptional effects of cancer-associated genomic aberrations

While the tumorigenic effects of specific recurrent mutations in known cancer driver-genes is well-characterized, not much is known about the functional relevance of the vast majority of recurrent mutations observed across cancers. Prior studies have attempted to identify functional genomic aberrations by integrating multi-omics measurements in cancer samples with community-curated biological pathway networks. However, the majority of these approaches overlook the following biological considerations: i) signaling pathway networks are highly tissue-specific and their regulatory interactions differ across tissue types; ii) regulatory factors exhibit heterogeneous influence on downstream gene transcription; iii) epigenetic and genomic alterations exhibit nonlinear impact on gene transcription. In order to accommodate these biological effects, we propose a hybrid Bayesian method to learn tissue-specific pairwise influence models amongst genes and to predict a gene's expression level as a nonlinear-function of its epigenetic and regulatory influences. We employ a novel tree-based depth-penalization mechanism in order to capture the higher regulatory impact of closer neighbors in the regulatory network. Using a breast cancer multi-omics dataset (N=1190), we show that our proposed method has superior prediction power over optimization-based regression models, with the additional advantage of revealing gene deregulations potentially driven by somatic mutations.

A new correlation clustering method for cancer mutation analysis

MOTIVATION

Cancer genomes exhibit a large number of different alterations that affect many genes in a diverse manner. An improved understanding of the generative mechanisms behind the mutation rules and their influence on gene community behavior is of great importance for the study of cancer.

RESULTS

To expand our capability to analyze combinatorial patterns of cancer alterations, we developed a rigorous methodology for cancer mutation pattern discovery based on a new, constrained form of correlation clustering. Our new algorithm, named C³ (Cancer Correlation Clustering), leverages mutual exclusivity of mutations, patient coverage and driver network concentration principles. To test C³, we performed a detailed analysis on TCGA breast cancer and glioblastoma data and showed that our algorithm outperforms the state-of-the-art CoMEt method in terms of discovering mutually exclusive gene modules and identifying biologically relevant driver genes. The proposed agnostic clustering method represents a unique tool for efficient and reliable identification of mutation patterns and driver pathways in large-scale cancer genomics studies, and it may also be used for other clustering problems on biological graphs.

AVAILABILITY AND IMPLEMENTATION

The source code for the C³ method can be found at https://github.com/jackhou2/C3 CONTACTS: jianma@cs.cmu.edu or milenkov@illinois.eduSupplementary information: Supplementary data are available at Bioinformatics online.

Hotspot mutations delineating diverse mutational signatures and biological utilities across cancer types

Background

An important step towards personalizing cancer treatment is to integrate heterogeneous evidences to catalog mutational hotspots that are biologically and therapeutically relevant and thus represent where targeted therapy would likely be beneficial. However, existing methods do not sufficiently delineate varying functionality of individual mutations within the same genes.

Results

We observed a large discordancy of mutation rates across different mutation subtypes and tumor types, and nominated 702 hotspot mutations in 549 genes in the Catalog of Somatic Mutations in Cancer (COSMIC) by considering context specific mutation characteristics such as genes, cancer types, mutation rates, mutation subtypes and sequence contexts. We observed that hotspot mutations were highly prevalent in Non CpG-island C/G transition and transversion sequence contexts in 10 tumor types, and specific insertion hotspot mutations were enriched in breast cancer and deletion hotspot mutations in colorectal cancer. We found that the hotspot mutations nominated by our approach were significantly more conserved than non-hotspot mutations in the corresponding cancer genes. We also examined the biological significance and pharmacogenomics properties of these hotspot mutations using data in the Cancer Genome Atlas (TCGA) and the Cancer Cell-Line Encyclopedia (CCLE), and found that 53 hotspot mutations are independently associated with diverse functional evidences in 1) mRNA and protein expression, 2) pathway activity, or 3) drug sensitivity and 82 were highly enriched in specific tumor types. We highlighted the distinct functional indications of hotspot mutations under different contexts and nominated novel hotspot mutations such as MAP3K4 A1199 deletion, NR1H2 Q175 insertion, and GATA3 P409 insertion as potential biomarkers or drug targets.

Conclusion

We identified a set of hotspot mutations across 17 tumor types by considering the background mutation rate variations among genes, tumor subtypes, mutation subtypes, and sequence contexts. We illustrated the common and distinct mutational signatures of hotspot mutations among different tumor types and investigated their variable functional relevance under different contexts, which could potentially serve as a resource for explicitly selecting targets for diagnosis, drug development, and patient management.

Electronic supplementary material

The online version of this article (doi:10.1186/s12864-016-2727-x) contains supplementary material, which is available to authorized users.

Identifying Network Perturbation in Cancer

We present a computational framework, called DISCERN (DIfferential SparsE Regulatory Network), to identify informative topological changes in gene-regulator dependence networks inferred on the basis of mRNA expression datasets within distinct biological states. DISCERN takes two expression datasets as input: an expression dataset of diseased tissues from patients with a disease of interest and another expression dataset from matching normal tissues. DISCERN estimates the extent to which each gene is perturbed-having distinct regulator connectivity in the inferred gene-regulator dependencies between the disease and normal conditions. This approach has distinct advantages over existing methods. First, DISCERN infers conditional dependencies between candidate regulators and genes, where conditional dependence relationships discriminate the evidence for direct interactions from indirect interactions more precisely than pairwise correlation. Second, DISCERN uses a new likelihood-based scoring function to alleviate concerns about accuracy of the specific edges inferred in a particular network. DISCERN identifies perturbed genes more accurately in synthetic data than existing methods to identify perturbed genes between distinct states. In expression datasets from patients with acute myeloid leukemia (AML), breast cancer and lung cancer, genes with high DISCERN scores in each cancer are enriched for known tumor drivers, genes associated with the biological processes known to be important in the disease, and genes associated with patient prognosis, in the respective cancer. Finally, we show that DISCERN can uncover potential mechanisms underlying network perturbation by explaining observed epigenomic activity patterns in cancer and normal tissue types more accurately than alternative methods, based on the available epigenomic data from the ENCODE project.

Translation of Genotype to Phenotype by a Hierarchy of Cell Subsystems

Accurately translating genotype to phenotype requires accounting for the functional impact of genetic variation at many biological scales. Here we present a strategy for genotype-phenotype reasoning based on existing knowledge of cellular subsystems. These subsystems and their hierarchical organization are defined by the Gene Ontology or a complementary ontology inferred directly from previously published datasets. Guided by the ontology's hierarchical structure, we organize genotype data into an "ontotype," that is, a hierarchy of perturbations representing the effects of genetic variation at multiple cellular scales. The ontotype is then interpreted using logical rules generated by machine learning to predict phenotype. This approach substantially outperforms previous, non-hierarchical methods for translating yeast genotype to cell growth phenotype, and it accurately predicts the growth outcomes of two new screens of 2,503 double gene knockouts impacting DNA repair or nuclear lumen. Ontotypes also generalize to larger knockout combinations, setting the stage for interpreting the complex genetics of disease.

Integrative modeling of multi-omics data to identify cancer drivers and infer patient-specific gene activity

Background

High throughput technologies have been used to profile genes in multiple different dimensions, such as genetic variation, copy number, gene and protein expression, epigenetics, metabolomics. Computational analyses often treat these different data types as independent, leading to an explosion in the number of features making studies under-powered and more importantly do not provide a comprehensive view of the gene’s state. We sought to infer gene activity by integrating different dimensions using biological knowledge of oncogenes and tumor suppressors.

Results

This paper proposes an integrative model of oncogene and tumor suppressor activity in cells which is used to identify cancer drivers and compute patient-specific gene activity scores. We have developed a Fuzzy Logic Modeling (FLM) framework to incorporate biological knowledge with multi-omics data such as somatic mutation, gene expression and copy number measurements. The advantage of using a fuzzy logic approach is to abstract meaningful biological rules from low-level numerical data. Biological knowledge is often qualitative, thus combining it with quantitative numerical measurements may leverage new biological insights about a gene’s state. We show that the oncogenic and altered tumor suppressing state of a gene can be better characterized by integrating different molecular measurements with biological knowledge than by each data type alone. We validate the gene activity score using data from the Cancer Cell Line Encyclopedia and drug sensitivity data for five compounds: BYL719 (PIK3CA inhibitor), PLX4720 (BRAF inhibitor), AZD6244 (MEK inhibitor), Erlotinib (EGFR inhibitor), and Nutlin-3 (MDM2 inhibitor). The integrative score improves prediction of drug sensitivity for the known drug targets of these compounds compared to each data type alone. The gene activity scores are also used to cluster colorectal cancer cell lines. Two subtypes of CRCs were found and potential cancer drivers and therapeutic targets for each of the subtypes were identified.

Conclusions

We propose a fuzzy logic based approach to infer gene activity in cancer by integrating numerical data with descriptive biological knowledge. We compute general patient-specific gene-level scores useful to determine the oncogenic or tumor suppressor status of cancer gene drivers and to cluster or classify patients.

Electronic supplementary material

The online version of this article (doi:10.1186/s12918-016-0260-9) contains supplementary material, which is available to authorized users.

Integrating mutation and gene expression cross-sectional data to infer cancer progression

Background

A major problem in identifying the best therapeutic targets for cancer is the molecular heterogeneity of the disease. Cancer is often caused by an accumulation of mutations which produce irreversible damage to the cell’s control mechanisms of survival and proliferation. Different mutations may affect these cellular anachronisms through a combination of molecular interactions which may be dynamically changing during cancer progression. It has been previously shown that cancer accumulates mutations over time. In this paper we address the problem of cancer heterogeneity by modeling cancer progression using somatic mutation and gene expression cross-sectional data.

Results

We propose a novel formulation of integrating somatic mutation and gene expression data to infer the temporal sequence of events from cross-sectional data. Using a mixed integer linear program we model the interaction between groups of different mutated genes and the resulting modifications at the gene expression level. Our approach identifies a partition of mutation events which gradually produce gene expression changes to a partition of genes over time. The proposed formulation is tested using both simulated data and real breast cancer data with matched somatic mutations and gene expression measurements from The Cancer Genome Atlas. First, we classify the genes as oncogenes or tumor suppressors based on the frequency of driver mutations. As expected, the most frequently mutated genes in breast cancer are PIK3CA and TP53 genes. Then, we select those genes with most frequent driver mutations and a set of genes known to play roles in cancer development. Furthermore, we apply the proposed mixed integer linear program to identify the temporal order in which genes mutate and, simultaneously, the changes they produce at the gene expression level during cancer progression. In addition, we are able to identify known causal relationships between mutations and gene expression changes in PI3K/AKT and TP53 pathways.

Conclusions

This paper proposes a new model to infer the temporal sequence in which mutations occur and lead to changes at the gene expression level during cancer progression. The approach is general and can be applied to any data sets with available somatic mutations and gene expression measurements.

Electronic supplementary material

The online version of this article (doi:10.1186/s12918-016-0255-6) contains supplementary material, which is available to authorized users.

Providing data science support for systems pharmacology and its implications to drug discovery

INTRODUCTION

The conventional one-drug-one-target-one-disease drug discovery process has been less successful in tracking multi-genic, multi-faceted complex diseases. Systems pharmacology has emerged as a new discipline to tackle the current challenges in drug discovery. The goal of systems pharmacology is to transform huge, heterogeneous, and dynamic biological and clinical data into interpretable and actionable mechanistic models for decision making in drug discovery and patient treatment. Thus, big data technology and data science will play an essential role in systems pharmacology.

AREAS COVERED

This paper critically reviews the impact of three fundamental concepts of data science on systems pharmacology: similarity inference, overfitting avoidance, and disentangling causality from correlation. The authors then discuss recent advances and future directions in applying the three concepts of data science to drug discovery, with a focus on proteome-wide context-specific quantitative drug target deconvolution and personalized adverse drug reaction prediction.

EXPERT OPINION

Data science will facilitate reducing the complexity of systems pharmacology modeling, detecting hidden correlations between complex data sets, and distinguishing causation from correlation. The power of data science can only be fully realized when integrated with mechanism-based multi-scale modeling that explicitly takes into account the hierarchical organization of biological systems from nucleic acid to proteins, to molecular interaction networks, to cells, to tissues, to patients, and to populations.

Discovering potential cancer driver genes by an integrated network-based approach

An integrated network-based approach is proposed to nominate driver genes. It is composed of two steps including a network diffusion step and an aggregated ranking step, which fuses the correlation between the gene mutations and gene expression, the relationship between the mutated genes and the heterogeneous characteristic of the patient mutation.

Identification of potential COPD genes based on multi-omics data at the functional level

Chronic obstructive pulmonary disease (COPD) is a complex disease, which involves dysfunctions in multi-omics. The changes in biological processes, such as adhesion junction, signaling transduction, transcriptional regulation, and cell proliferation, will lead to the occurrence of COPD. A novel systematic approach MMMG (Methylation-MicroRNA-MRNA-GO) was proposed to identify potential COPD genes by integrating function information with a methylation profile, a microRNA expression profile and an mRNA expression profile. 8 co-functional classes and 102 potential COPD genes were identified. These genes displayed a high performance in classifying COPD patients and normal samples, revealed COPD-related pathways, and have been confirmed to be associated with COPD by Matthews correlation coefficient (MCC)-values, literature, an independent data set, and pathways. The MMMG method that analyzed multi-omics data at the functional level could effectively identify potential COPD genes. These potential COPD genes would provide in-depth insights into understanding the complexity of COPD genome landscapes, improve the early diagnostics, and guide new efforts to develop therapeutics in the future.

HMMvar-func: a new method for predicting the functional outcome of genetic variants

Background

Numerous tools have been developed to predict the fitness effects (i.e., neutral, deleterious, or beneficial) of genetic variants on corresponding proteins. However, prediction in terms of whether a variant causes the variant bearing protein to lose the original function or gain new function is also needed for better understanding of how the variant contributes to disease/cancer. To address this problem, the present work introduces and computationally defines four types of functional outcome of a variant: gain, loss, switch, and conservation of function. The deployment of multiple hidden Markov models is proposed to computationally classify mutations by the four functional impact types.

Results

The functional outcome is predicted for over a hundred thyroid stimulating hormone receptor (TSHR) mutations, as well as cancer related mutations in oncogenes or tumor suppressor genes. The results show that the proposed computational method is effective in fine grained prediction of the functional outcome of a mutation, and can be used to help elucidate the molecular mechanism of disease/cancer causing mutations. The program is freely available at http://bioinformatics.cs.vt.edu/zhanglab/HMMvar/download.php.

Conclusion

This work is the first to computationally define and predict functional impact of mutations, loss, switch, gain, or conservation of function. These fine grained predictions can be especially useful for identifying mutations that cause or are linked to cancer.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-015-0781-z) contains supplementary material, which is available to authorized users.

Pathway and network analysis of cancer genomes

Genomic information on tumors from 50 cancer types cataloged by the International Cancer Genome Consortium (ICGC) shows that only a few well-studied driver genes are frequently mutated, in contrast to many infrequently mutated genes that may also contribute to tumor biology. Hence there has been large interest in developing pathway and network analysis methods that group genes and illuminate the processes involved. We provide an overview of these analysis techniques and show where they guide mechanistic and translational investigations.