All cancers are caused by somatic mutations. These mutations may be the consequence of the intrinsic slight infidelity of the DNA replication machinery, exogenous or endogenous mutagen exposures, enzymatic modification of DNA, or defective DNA repair. In some cancer types, a substantial proportion of somatic mutations are known to be generated by exogenous carcinogens, for example, tobacco smoking in lung cancers and ultraviolet light in skin cancers, or by abnormalities of DNA maintenance, for example, defective DNA mismatch repair in some colorectal cancers. Each biological process causing mutations leaves a characteristic imprint on the genome of a cancer cell, termed, mutational signature. In this talk, I will present mutational signatures analyses encompassing 12,023 cancer genomes across 40 distinct types of human cancer revealing more than 30 different signatures of mutational processes. Some signatures are present in many cancer types, notably a signature attributed to the APOBEC family of cytidine deaminases, whereas others are confined to a single cancer class. Certain signatures are associated with age of the patient at cancer diagnosis, known mutagenic exposures or defects in DNA maintenance, but many are of cryptic origin. The results reveal the diversity of mutational processes underlying the development of cancer, with potential implications for understanding of cancer etiology, prevention and therapy.

My presentation will cover research from our group on polymorphic genome structural variation, and on genomic somatic DNA rearrangements in cancer. I will provide an update on our efforts to reconstruct patterns of polymorphic genome structural variation through analysis of DNA sequencing data from the 1000 Genomes Project and its follow-up study on structural variation, the Human Genome Structural Variation Consortium. Furthermore, we are in the process of using approaches developed in our group to perform an analysis of >2,800 deeply sequenced tumor/normal paired genomes in the context of the Pan Cancer Analysis of Whole Genomes (PCAWG) project, to search for commonalities and differences in molecular processes leading to cancer in different tumor entities. Based on pan-cancer analysis we recently inferred, and subsequently verified experimentally, a novel mechanism of cancer gene activation involving somatic de novo 3D contact domain formation mediated by recurrent tandem duplications, which activate the IGF2 oncogene locus. We further have recently developed a novel experimental approach to recapitulate a stunning structural alteration mechanism, termed chromothripsis, which involves massive DNA rearrangements that can reshuffle the cancer genome in a single cell division cycle. These results illustrate approaches used by our group for uncovering determinants behind the formation and selection of genetic variation in the human genome.

Immune cells form a coordinated hierarchy to adapt to genetic and environmental contexts. Immune responses involve coordination across cell types and tissues. We have recently developed experimental and analytical methods for assessing the state of the immune system across the organism, enabling detailed and comprehensive studies of immune responses in various contexts. I will discuss recent and ongoing efforts to model anti-tumor immune responses in the context of cancer immunotherapy, which have revealed an essential role for systemic immunity in mediating tumor rejection.

Dysregulated epigenetic developmental programs are a feature of many cancers, and the diverse differentiation states of immune cells as well as their dysfunctional states in tumors are in part epigenetically encoded.  We developed an integrative computational strategy to exploit genome-wide data on chromatin accessibility (DNase-seq or ATAC-seq), histone modifications (ChIP-seq), and transcription (RNA-seq) in order to study enhancer landscape and gene expression dynamics in cellular differentiation, with a focus on the hematopoietic system.  We examined how early establishment of enhancers and complex regulatory locus control together govern gene expression changes in cell state transitions. We also developed a quantitative model to predict gene expression changes from the DNA sequence content and lineage history of active enhancers.  We are now using these methods to study the chromatin dynamics of neoantigen-specific T cells in tumors.  It was previously found that tumor-specific CD8 T cells (TSTs) differentiate to dysfunctionality through two discrete functional states: an initial plastic state that can be functionally rescued and a later fixed state that is resistant to therapeutic reprogramming. We profiled the gene expression (RNA-seq) and chromatin accessibility (ATAC-seq) landscapes of functional T cells and dysfunctional TSTs using a mouse genetic system where the SV40 T antigen serves as both the oncogenic driver and the neoantigen.  We performed integrative computational analysis and follow-up experiments to show that plastic and fixed dysfunction are encoded by distinct epigenetic states, to identify transcription factors (TFs) with global changes in binding site accessibility during progression to the fixed dysfunctional state, and to show that in vivo pharmacological modulation of identified TFs decreases or delays dysfunction.  We also showed that patient-derived PD1 high tumor infiltrating lymphocytes (TILs) display an epigenetic signature of fixed dysfunction, and we defined novel cell surface markers that demarcated reprogrammable TSTs within the heterogeneous TIL population, a finding of potential clinical significance.

One of the main challenges in precision medicine is to predict whether a patient will benefit from a given therapy. In this context, it is crucial to develop computational models predictive of therapy response based on molecular features from diseased tissues. Such models could be used to assist clinicians in their decision process and consequently improve disease management. However, testing the effect of drugs in patient is lengthy and costly. Large-scale drug screening in cell lines (in vitro) has recently been used to generate large amount of molecular data related to drug response. These pharmacogenomic studies hold great promises in identifying biomarkers predictive of therapy response in model systems, which could be further translated into clinical settings. However, discovery of predictive biomarkers in vitro has proven to be extremely challenging. We recently showed that this could be due to inconsistencies in drug response data, leading to irreproducible gene-drug associations across datasets [1]. However, our findings originating from the comparison of two independent large-scale pharmacogenomic screens, Genomics of Drug Sensitivity in Cancer (GDSC) [1] and Cancer Cell Line Encyclopedia (CCLE) [2], were recently contradicted by the GDSC and CCLE investigators who recently reported reasonable agreement of pharmacological phenotypes and molecular predictors of drug response [4]. Reanalyzing the authors’ published methods and results, we found that their analysis failed to account for variability in the genomic data and more importantly compared different drug sensitivity measures from each study, which substantially deviates from our more stringent consistency assessment. Another study from Genentech [5] compared the GDSC and CCLE data to their own datasets and support strong agreement between these datasets although they found the pharmacological assay used in GDSC to be less reliable. We are now updating our comparison of the most updated genomic and pharmacological data from these studies to shed light on the (in)consistency in pharmacogenomic studies [6]. Our experience illustrates the importance of data and code sharing in bioinformatics and biomedical research, as the data generation process becomes increasingly complex and requires high level of replication to achieve robust results.

Cell migration is key to many biological processes including embryonic development, wound healing, and disease progression. Recent studies have shown that cell migration is sensitive to microenvironmental stiffness, and many cells display a stiffness optimum at which migration is maximal. To create a unified theoretical framework for cell migration, we have now developed and experimentally tested a whole cell migration simulator based on the motor-clutch model of cellular force transmission by imposing coupled force balances and mass balances on molecular motors, adhesion molecules (“clutches”), and actin subunits in a compliant microenvironment. The model predicts a stiffness optimum that can be shifted by altering the number of active molecular motors and clutches. This prediction was tested experimentally, and we find that the motor-clutch cell migration simulator provides a theoretical framework with which to predict cell adhesion and migration in defined mechanochemical microenvironments in 1D, 2D, and 3D.

Circulating tumor DNA (ctDNA) represents a promising biomarker for sensitive, specific, and dynamic detection of disease burden in cancer patients. Mutations in tumor-derived DNA represent ideal potential biomarkers since they are highly specific to tumor cells and involved in disease pathogenesis. However, even in advanced cancer patients concentrations of ctDNA are often low and difficult to detect. We have developed a novel, ultra-sensitive and specific method for detection of circulating tumor DNA called Cancer Personalized Profiling by Deep Sequencing (CAPP-Seq). This method was developed specifically for detection of ctDNA in non-small cell lung cancer patients, although it is broadly applicable to other cancer types. In this presentation I will describe our recent work on applications of ctDNA analysis in a variety of clinical settings.

Previous studies have revealed extraordinary diversity in cancer genomes through the analysis of bulk tumor biopsies, however large gaps remain in our understanding of tumor cellular composition. Although single cell RNA sequencing has recently emerged as a powerful technology for defining novel cell subsets, it is currently impractical for large-scale analyses and cannot be applied to fixed specimens collected as part of routine clinical care. To facilitate profiling of cellular heterogeneity in fresh, frozen, and fixed tissues, we developed an “in silico cytometry” method for characterizing cell subsets of interest from gene expression profiles of intact bulk tumors. In this presentation, I will describe this framework and its applications for understanding neoplastic tissue composition.

A central question in cancer immunity is what drives effective immunity against cancer in the context of immunotherapy or spontaneously in patients. Using a variety of methods, we have identified tumor antigens, immune cell types and states and evolutionary genetic changes in tumors, which explain some of the success and failure of anti-tumor immunity in humans. We have also used some of these results to develop fully personalized therapeutics that target tumors in patients.