Search

The growth of multi-omic tumour profile datasets along with knowledge of genome regulatory networks has created an unprecedented opportunity to advance precision oncology. Achieving this goal requires computational methods that can make sense of and combine heterogeneous data sources. Interpretability and integration of prior knowledge is of particular relevance for genomic models to minimize ungeneralizable models, promote rational treatment design, and make use of sparse genetic mutation data. While networks have long been used to capture genomic interactions at the levels of genes, proteins, and pathways, the use of networks in precision oncology is relatively new. In this chapter, I provide an introduction to network-based approaches used to integrate multi-modal data sources for patient stratification and patient classification. There is a particular emphasis on methods using patient similarity networks (PSNs) as part of the design. I separately discuss strategies for inferring driver mutations from individual patient mutation data. Finally, I discuss challenges and opportunities the field will need to overcome to achieve its full potential, with an outlook towards a clinic of the future.

Clinical research and practice in the 21(st) century is poised to be transformed by analysis of computable electronic medical records and population-level genome-scale patient profiles. Genomic data captures genetic and environmental state, providing information about heterogeneity in disease and treatment outcome, but genomic-based clinical risk scores are limited. Achieving the goal of routine precision medicine that takes advantage of this rich genomics data will require computational methods that support heterogeneous data, have excellent predictive performance, and ideally, provide biologically-interpretable results. Traditional machine-learning approaches excel at performance, but often have limited interpretability. Patient similarity networks are an emerging paradigm for precision medicine, in which patients are clustered or classified based on their similarities in various features, including genomic profiles. This strategy is analogous to standard medical diagnosis, has excellent performance, is interpretable, and can preserve patient privacy. We review new methods based on patient similarity networks, including Similarity Network Fusion for patient clustering and netDx for patient classification. While these methods are already useful, much work is required to improve their scalability for contemporary genetic cohorts, optimize parameters, and incorporate a wide range of genomics and clinical data. The coming five years will provide an opportunity to assess the utility of network-based algorithms for precision medicine.

An open challenge in human genetics is to better understand the link between genotype variation and the various molecular, cellular, anatomical and physiological systems that it can affect. To address this challenge, we performed genotype-phenotype-systems analysis for accuracy in nine cognitive tasks from the Philadelphia Neurodevelopmental Cohort (3,319 individuals aged 8-21 years). We report a region of genome-wide significance within the 3’ end of the FBLN1 gene (p=4.6×10−8), associated with nonverbal reasoning, a heritable form of complex reasoning ability. Integration of published brain-specific ‘omic maps reveals that FBLN1 shows greatest expression in the fetal brain, is a marker of neural progenitor cells, is differentially expressed in schizophrenia and increases genetic risk for bipolar disorder. These findings suggest that nonverbal reasoning and FBLN1 variation warrant further investigation in studies of neurodevelopmental disorders and psychosis. Using genotype-pathway analysis, we identify pathways related to development and to autonomic nervous system dysfunction associated with working memory accuracy. Top-ranking pathway genes include those genetically associated with multiple diseases with working memory deficits, such as schizophrenia and Parkinson’s disease, and that are also markers for specific brain cell types. Our findings identify novel molecular players involved in specific cognitive tasks and link variants to genes, pathways, cell types, diseases and drugs. This work advances the “molecules-to-behaviour” view of cognition, and provides a framework for using systems-level organization of data for other biomedical domains.

Patient classification has widespread biomedical and clinical applications, including diagnosis, prognosis and treatment response prediction. A clinically useful prediction algorithm should be accurate, generalizable, be able to integrate diverse data types, and handle sparse data. A clinical predictor based on genomic data needs to be easily interpretable to drive hypothesis-driven research into new treatments. We describe netDx, a novel supervised patient classification framework based on patient similarity networks. netDx meets the above criteria and particularly excels at data integration and model interpretability. As a machine learning method, netDx demonstrates consistently excellent performance in a cancer survival benchmark across four cancer types by integrating up to six genomic and clinical data types. In these tests, netDx has significantly higher average performance than most other machine-learning approaches across most cancer types and its best model outperforms all other methods for two cancer types. In comparison to traditional machine learning-based patient classifiers, netDx results are more interpretable, visualizing the decision boundary in the context of patient similarity space. When patient similarity is defined by pathway-level gene expression, netDx identifies biological pathways important for outcome prediction, as demonstrated in diverse data sets of breast cancer and asthma. Thus, netDx can serve both as a patient classifier and as a tool for discovery of biological features characteristic of disease. We provide a software complete implementation of netDx along with sample files and automation workflows in R.

Patient classification based on clinical and genomic data will further the goal of precision medicine. Interpretability is of particular relevance for models based on genomic data, where sample sizes are relatively small (in the hundreds), increasing overfitting risk netDx is a machine learning method to integrate multi-modal patient data and build a patient classifier. Patient data are converted into networks of patient similarity, which is intuitive to clinicians who also use patient similarity for medical diagnosis. Features passing selection are integrated, and new patients are assigned to the class with the greatest profile similarity. netDx has excellent performance, outperforming most machine-learning methods in binary cancer survival prediction. It handles missing data – a common problem in real-world data – without requiring imputation. netDx also has excellent interpretability, with native support to group genes into pathways for mechanistic insight into predictive features. The netDx Bioconductor package provides multiple workflows for users to build custom patient classifiers. It provides turnkey functions for one-step predictor generation from multi-modal data, including feature selection over multiple train/test data splits. Workflows offer versatility with custom feature design, choice of similarity metric; speed is improved by parallel execution. Built-in functions and examples allow users to compute model performance metrics such as AUROC, AUPR, and accuracy. netDx uses RCy3 to visualize top-scoring pathways and the final integrated patient network in Cytoscape. Advanced users can build more complex predictor designs with functional building blocks used in the default design. Finally, the netDx Bioconductor package provides a novel workflow for pathway-based patient classification from sparse genetic data.

A key challenge in the application of whole-genome sequencing (WGS) for clinical diagnostic and research is the high-throughput prioritization of functional variants in the non-coding genome. This challenge is compounded by context-specific genetic modulation of gene expression, and variant-gene mapping depends on the tissues and organ systems affected in a given disease; for instance, a disease affecting the gastrointestinal system would use maps specific to genome regulation in gut-related tissues. While there are large-scale atlases of genome regulation, such as GTEx and NIH Roadmap Epigenomics, the clinical genetics community lacks publicly-available stand-alone software for high-throughput annotation of custom variant data with user-defined tissue-specific epigenetic maps and clinical genetic databases, to prioritize variants for a specific biomedical application. In this work, we provide a simple software pipeline, called SNPnotes, which takes as input variant calls for a patient and prioritizes those using information on clinical relevance from ClinVar, tissue-specific gene regulation from GTEx and disease associations from the NHGRI-EBI GWAS catalogue. This pipeline was developed as part of SVAI Research's "Undiagnosed-1" event for collaborative patient diagnosis. We applied this pipeline to WGS-based variant calls for an individual with a history of gastrointestinal symptoms, using 12 gut-specific eQTL maps and GWAS associations for metabolic diseases, for variant-gene mapping. Out of 6,248,584 SNPs, the pipeline identified 151 high-priority variants, overlapping 129 genes. These top SNPs all have known clinical pathogenicity, modulate gene expression in gut tissues and have genetic associations with metabolic disorders, and serve as starting points for hypotheses about mechanisms driving clinical symptoms. Simple software changes can be made to customize the pipeline for other tissue-specific applications. Future extensions could integrate maps of tissue-specific regulatory elements, higher-order chromatin loops, and mutations affecting splice variants.

WormBase (http://www.wormbase.org), the model organism database for information about Caenorhabditis elegans and related nematodes, continues to expand in breadth and depth. Over the past year, WormBase has added multiple large-scale datasets including SAGE, interactome, 3D protein structure datasets and NCBI KOGs. To accommodate this growth, the International WormBase Consortium has improved the user interface by adding new features to aid in navigation, visualization of large-scale datasets, advanced searching and data mining. Internally, we have restructured the database models to rationalize the representation of genes and to prepare the system to accept the genome sequences of three additional Caenorhabditis species over the coming year.

Dynamic patterns of cytosine-5 methylation and successive hydroxylation are part of epigenetic regulation in eukaryotes, including humans, which contributes to normal phenotypic variation and disease risk. Here we present an approach for the mapping of unmodified regions of the genome, which we call the unmethylome. Our technique is based on DNA methyltransferase-directed transfer of activated groups and covalent biotin tagging of unmodified CpG sites followed by affinity enrichment and interrogation on tiling microarrays or next generation sequencing. Control experiments and pilot studies of human genomic DNA from cultured cells and tissues demonstrate that, along with providing a unique cross-section through the chemical landscape of the epigenome, the methyltransferase-directed transfer of activated groups-based approach offers high precision and robustness as compared with existing affinity-based techniques.