Search

To make machine learning analyses in the life sciences more computationally reproducible, we propose standards based on data, model, and code publication, programming best practices, and workflow automation. By meeting these standards, the community of researchers applying machine learning methods in the life sciences can ensure that their analyses are worthy of trust.

Low-coverage or shallow whole genome sequencing (sWGS) approaches can efficiently detect somatic copy number aberrations (SCNAs) at low cost. This is clinically important for many cancers, in particular cancers with severe chromosomal instability (CIN) that frequently lack actionable point mutations and are characterised by poor disease outcome. Absolute copy number (ACN), measured in DNA copies per cancer cell, is required for meaningful comparisons between copy number states, but is challenging to estimate and in practice often requires manual curation. Using a total of 60 cancer cell lines, 148 patient-derived xenograft (PDX) and 142 clinical tissue samples, we evaluate the performance of available tools for obtaining ACN from sWGS. We provide a validated and refined tool called Rascal (relative to absolute copy number scaling) that provides improved fitting algorithms and enables interactive visualisation of copy number profiles. These approaches are highly applicable to both pre-clinical and translational research studies on SCNA-driven cancers and provide more robust ACN fits from sWGS data than currently available tools.

Summary Intra-tumor heterogeneity (ITH) is a mechanism of therapeutic resistance and therefore an important clinical challenge. However, the extent, origin, and drivers of ITH across cancer types are poorly understood. To address this, we extensively characterize ITH across whole-genome sequences of 2,658 cancer samples spanning 38 cancer types. Nearly all informative samples (95.1%) contain evidence of distinct subclonal expansions with frequent branching relationships between subclones. We observe positive selection of subclonal driver mutations across most cancer types and identify cancer type-specific subclonal patterns of driver gene mutations, fusions, structural variants, and copy number alterations as well as dynamic changes in mutational processes between subclonal expansions. Our results underline the importance of ITH and its drivers in tumor evolution and provide a pan-cancer resource of comprehensively annotated subclonal events from whole-genome sequencing data., Graphical abstract, Highlights • Pan-cancer resource of comprehensively annotated intra-tumor heterogeneity (ITH) • ITH is pervasive across cancers and shows cancer type-specific patterns • Branching phylogenies are common • Dynamic changes in mutational processes between subclonal expansions, Dentro et al. provide a comprehensive annotation of intra-tumor heterogeneity and its drivers in cancer evolution.

Intra-tumor heterogeneity (ITH) is a mechanism of therapeutic resistance and therefore an important clinical challenge. However, the extent, origin and drivers of ITH across cancer types are poorly understood. To address this question, we extensively characterize ITH across whole-genome sequences of 2,658 cancer samples, spanning 38 cancer types. Nearly all informative samples (95.1%) contain evidence of distinct subclonal expansions, with frequent branching relationships between subclones. We observe positive selection of subclonal driver mutations across most cancer types, and identify cancer type specific subclonal patterns of driver gene mutations, fusions, structural variants and copy-number alterations, as well as dynamic changes in mutational processes between subclonal expansions. Our results underline the importance of ITH and its drivers in tumor evolution, and provide an unprecedented pan-cancer resource of comprehensively annotated subclonal events from whole-genome sequencing data.

BackgroundVirtUaL ChIP-seq Analysis through Networks (VULCAN) infers regulatory interactions of transcription factors by overlaying networks generated from publicly available tumor expression data onto ChIP-seq data. We apply our method to dissect the regulation of estrogen receptor-alpha activation in breast cancer to identify potential co-regulators of the estrogen receptor’s transcriptional response.ResultsVULCAN analysis of estrogen receptor activation in breast cancer highlights the key components of the estrogen receptor complex alongside a novel interaction with GRHL2. We demonstrate that GRHL2 is recruited to a subset of estrogen receptor binding sites and regulates transcriptional output, as evidenced by changes in estrogen receptor-associated eRNA expression and stronger estrogen receptor binding at active enhancers after GRHL2 knockdown.ConclusionsOur findings provide new insight into the role of GRHL2 in regulating eRNA transcription as part of estrogen receptor signaling. These results demonstrate VULCAN, available from Bioconductor, as a powerful predictive tool.

Recent advances in single-cell RNA-sequencing (scRNA-seq) in combination with CRISPR/Cas9 technologies have enabled the development of methods for large-scale perturbation studies with transcriptional readouts. These methods are highly scalable and have the potential to provide a wealth of information on the biological networks that underlie cellular response. Here we discuss how to overcome several key challenges to generate and analyse data for the confident reconstruction of models of the underlying cellular network. Some challenges are generic, and apply to analysing any single-cell transcriptomic data, while others are specific to combined single-cell CRISPR/Cas9 data, in particular barcode swapping, knockdown efficiency, multiplicity of infection and potential confounding factors. We also provide a curated collection of published data sets to aid the development of analysis strategies. Finally, we discuss several network reconstruction approaches, including co-expression networks and Bayesian networks, as well as their limitations, and highlight the potential of Nested Effects Models for network reconstruction from scRNA-seq data. This article is part of a Special Issue entitled: Transcriptional Profiles and Regulatory Gene Networks edited by Dr. Dr. Federico Manuel Giorgi and Dr. Shaun Mahony.

Estimating and clustering cancer cell fractions of genomic alterations are central tasks for studying intratumour heterogeneity. We present Ccube, a probabilistic framework for inferring the cancer cell fraction of somatic point mutations and the subclonal composition from whole-genome sequencing data. We develop a variational inference method for model fitting, which allows us to handle samples with large number of the variants (more than 2 million) while quantifying uncertainty in a Bayesian fashion. Ccube is available at https://github.com/keyuan/ccube.

Understanding the dynamics of endogenous protein–protein interactions in complex networks is pivotal in deciphering disease mechanisms. To enable the in-depth analysis of protein interactions in chromatin-associated protein complexes, we have previously developed a method termed RIME (Rapid Immunoprecipitation Mass spectrometry of Endogenous proteins). Here, we present a quantitative multiplexed method (qPLEX-RIME), which integrates RIME with isobaric labelling and tribrid mass spectrometry for the study of protein interactome dynamics in a quantitative fashion with increased sensitivity. Using the qPLEX-RIME method, we delineate the temporal changes of the Estrogen Receptor alpha (ERα) interactome in breast cancer cells treated with 4-hydroxytamoxifen. Furthermore, we identify endogenous ERα-associated proteins in human Patient-Derived Xenograft tumours and in primary human breast cancer clinical tissue. Our results demonstrate that the combination of RIME with isobaric labelling offers a powerful tool for the in-depth and quantitative characterisation of protein interactome dynamics, which is applicable to clinical samples., Chromatin-associated protein complexes play a critical role in the regulation of gene expression in health and disease. Here, the authors describe a sensitive mass spectrometry-based method to monitor the dynamic interactions of endogenous chromatin-associated protein complexes in clinical samples.

SUMMARYIntra-tumor heterogeneity (ITH) is a mechanism of therapeutic resistance and therefore an important clinical challenge. However, the extent, origin and drivers of ITH across cancer types are poorly understood. To address this question, we extensively characterize ITH across whole-genome sequences of 2,658 cancer samples, spanning 38 cancer types. Nearly all informative samples (95.1%) contain evidence of distinct subclonal expansions, with frequent branching relationships between subclones. We observe positive selection of subclonal driver mutations across most cancer types, and identify cancer type specific subclonal patterns of driver gene mutations, fusions, structural variants and copy-number alterations, as well as dynamic changes in mutational processes between subclonal expansions. Our results underline the importance of ITH and its drivers in tumor evolution, and provide an unprecedented pan-cancer resource of comprehensively annotated subclonal events from whole-genome sequencing data.

Background VirtUaL ChIP-seq Analysis through Networks (VULCAN) infers regulatory interactions of transcription factors by overlaying networks generated from publicly available tumor expression data onto ChIP-seq data. We applied our method to dissect the regulation of Estrogen Receptor-alpha (ER) activation in breast cancer to identify potential coregulators of the ER’s transcriptional response. Results VULCAN analysis of ER activation in breast cancer highlighted key components of the ER complex alongside a novel interaction with GRHL2. We demonstrate that GRHL2 is recruited to a subset of ER binding sites and regulates the transcriptional output of ER, as evidenced by changes in ER-associated eRNA expression, and stronger ER binding at active enhancers (H3K27ac sites) after GRHL2 knockdown. Conclusions Our findings provide new insight into the role of GRHL2 in regulating eRNA transcription as part of ER signaling. These results demonstrate VULCAN, available from Bioconductor, as a powerful predictive tool.

Tumours with profound copy-number aberration elude molecular stratification due to their genomic complexity. By representing this complexity as a mixture of copy-number signatures, we provide molecular explanations for differing clinical outcomes. Here we present a method for copy-number signature identification, deriving eight signatures in 117 shallow whole-genome sequenced high-grade serous ovarian cancers (HGSOC), which validated on independent cohorts of 95 deep whole-genome sequenced, and 402 SNP array-profiled cases. Three copy-number signatures predicted longer overall survival, while the others predicted poorer outcome. We found evidence for the mutational processes giving rise to copy-number change for six of the eight signatures via correlations with other genomic features. Our results provide insights into the pathogenesis of HGSOC by uncovering multiple mutational processes that shape genomes following TP53 mutation. Importantly, our work shows that most HGSOC have a mixture of mutational processes suggesting that targeting a single mutator phenotype may be therapeutically suboptimal.

A key challenge in quantitative ChIP-seq is the normalisation of data in the presence of genome-wide changes in occupancy. Analysis-based normalisation methods were developed for transcriptomic data and these are dependent on the underlying assumption that total transcription does not change between conditions. For genome-wide changes in transcription factor binding, these assumptions do not hold true. The challenges in normalisation are confounded by experimental variability during sample preparation, processing, and recovery. We present a novel normalisation strategy utilising an internal standard of unchanged peaks for reference. Our method can be readily applied to monitor genome- wide changes by ChIP-seq that are otherwise lost or misrepresented through analytical normalisation. We compare our approach to normalisation by total read depth and two alternative methods that utilise external experimental controls to study transcription factor binding. We successfully resolve the key challenges in quantitative ChIP-seq analysis and demonstrate its application by monitoring the loss of Estrogen Receptor-alpha (ER) binding upon fulvestrant treatment, ER binding in response to estrodiol, ER mediated change in H4K12 acetylation and profiling ER binding in Patient-Derived Xenographs. This is supported by an adaptable pipeline to normalise and quantify differential transcription factor binding genome- wide and generate metrics for differential binding at individual sites.

We present SVclone, a computational method for inferring the cancer cell fraction of structural variant breakpoints from whole-genome sequencing data. We validate our approach using simulated and real tumour samples, and demonstrate its utility on 2,778 whole-genome sequenced tumours. We find a subset of liver, breast and ovarian cancer cases with decreased overall survival that have subclonally enriched copy-number neutral rearrangements, an observation that could not be discovered with currently available methods.

Elucidating interactions between cancer cells and their microenvironment is a key goal of cancer research with implications for understanding cancer evolution and improving immunotherapy. Previous studies used association-based approaches to infer relationships in transcriptomic data, but could not infer the direction of interaction. Here we present a causal modeling approach that infers directed interactions between signaling pathway activity and immune activity by anchoring the analysis on somatic genomic changes. Our approach integrates copy number profiles, transcriptomic data, image data and a protein-protein interaction network to infer directed relationships. As a result, we propose 11 novel genomic drivers of T cell phenotypes in the breast cancer tumour microenvironment and validate them in independent cohorts and orthogonal data types. Our framework is flexible and provides a generally applicable way to extend association-based analysis in other cancer types and to other data and clinical parameters.

Allele-specific measurements of transcription factor binding from ChIP-seq data are key to dissecting the allelic effects of non-coding variants and their contribution to phenotypic diversity. However, most methods of detecting an allelic imbalance assume diploid genomes. This assumption severely limits their applicability to cancer samples with frequent DNA copy-number changes. Here we present a Bayesian statistical approach called BaalChIP to correct for the effect of background allele frequency on the observed ChIP-seq read counts. BaalChIP allows the joint analysis of multiple ChIP-seq samples across a single variant and outperforms competing approaches in simulations. Using 548 ENCODE ChIP-seq and six targeted FAIRE-seq samples, we show that BaalChIP effectively corrects allele-specific analysis for copy-number variation and increases the power to detect putative cis-acting regulatory variants in cancer genomes. Electronic supplementary material The online version of this article (doi:10.1186/s13059-017-1165-7) contains supplementary material, which is available to authorized users.

PLoS Computational Biology, 13 (4), ISSN:1553-734X, ISSN:1553-7358

BACKGROUND: The genomic complexity of profound copy-number aberration has prevented effective molecular stratification of high grade serous ovarian carcinoma (HGSOC). Recent algorithmic advances have enabled interpretation of complex genomic changes by identifying mutational signatures—genomic patterns that are the imprint of mutagenic processes accumulated over the lifetime of a cancer cell. We hypothesized that specific features of copy-number (CN) abnormalities could represent the imprints of distinct mutational processes, and developed methods to identify signatures from copy-number features in HGSOC. METHODS: We derived copy-number signatures from absolute copy number profiles from 253 primary and relapsed HGSOC samples from 132 patients in the BriTROC-1 cohort using low-cost shallow whole-genome sequencing (sWGS; 0.1×). A subset of 56 of these cases had deep whole-genome sequencing (dWGS) performed for mutation analysis and comparison with sWGS data. Independent validation was performed using 112 dWGS HGSOC cases from PCAWG and 415 HGSOC cases with SNP array and whole exome sequence from TCGA. CN signature exposures were correlated with mutation data, SNV signatures, and other measures derived from deep WGS and exome sequencing to identify statistically significant genomic associations using a false discovery rate RESULTS: We identified 7 CN signatures that provided a molecular framework to rederive the major defining elements of HGSOC genomes, including defective homologous recombination (HRD), tandem duplication, amplification of CCNE1 and amplification-associated fold-back inversions. Almost all patients with HGSOC demonstrated a mixture of signatures indicative of combinations of mutational processes, including those with early driver events such as BRCA2 mutation (in addition to HRD signatures). High exposure to CN signature 3, characterised by BRCA1/2-related HRD, was associated with improved overall survival. Conversely, high exposure to signature 1, which was characterised by oncogenic RAS signaling (including NF1, KRAS and NRAS mutation), predicted platinum-resistant relapse and poor survival. CONCLUSIONS: HGSOC lacks clinically-relevant patient stratification, which is reflected in poor survival and is a significant barrier to precision medicine. Copy-number signature exposures at diagnosis predict both overall survival and the probability of platinum-resistant relapse. Our results suggest that early TP53 mutation, the ubiquitous initiating event in HGSOC, may permit multiple mutational processes to co-evolve, potentially simultaneously and that additional signature exposures may alter the risk of developing therapeutic resistance. Thus, our results suggest that HGSOC is a continuum of genomes. We derived signatures using inexpensive sWGS of DNA from core biopsies. These approaches are rapid and cost effective, thus providing a clear path to clinical implementation. By dissecting the mutational forces shaping HGSOC genomes, our study paves the way to understanding extreme genomic complexity, as well as revealing the evolution of tumors as they relapse and acquire resistance to therapy. Citation Format: Geoff Macintyre, Teodora E. Goranova, Dilrini De Silva, Darren Ennis, Anna M. Piskorz, Matthew Eldridge, Daoud Sie, Liz-Anne Lewsley, Aishah Hanif, Cheryl Wilson, Suzanne Dowson, Rosalind M. Glasspool, Michelle Lockley, Elly Brockbank, Ana Montes, Axel Walther, Sudha Sundar, Richard Edmondson, Geoff D. Hall, Andrew Clamp, Charlie Gourley, Marcia Hall, Christina Fotopoulou, Hani Gabra, James Paul, Anna Supernat, David Millan, Aoisha Hoyle, Gareth Bryson, Craig Nourse, Laura Mincarelli, Luis Navarro Sanchez, Bauke Ylstra, Mercedes Jimenez-Linan, Luiza Moore, Oliver Hofmann, Florian Markowetz, Iain A. McNeish, James D. Brenton. COPY-NUMBER SIGNATURES AND MUTATIONAL PROCESSES IN HIGH GRADE SEROUS OVARIAN CARCINOMA [abstract]. In: Proceedings of the 12th Biennial Ovarian Cancer Research Symposium; Sep 13-15, 2018; Seattle, WA. Philadelphia (PA): AACR; Clin Cancer Res 2019;25(22 Suppl):Abstract nr AP09.

Tumours develop in an evolutionary process, in which the accumulation of mutations produces subpopulations of cells with distinct mutational profiles, called clones. This process leads to the genetic heterogeneity widely observed in tumour sequencing data, but identifying the genotypes and frequencies of the different clones is still a major challenge. Here, we present Cloe, a phylogenetic latent feature model to deconvolute tumour sequencing data into a set of related genotypes. Our approach extends latent feature models by placing the features as nodes in a latent tree. The resulting model can capture both the acquisition and the loss of mutations, as well as episodes of convergent evolution. We establish the validity of Cloe on synthetic data and assess its performance on controlled biological data, comparing our reconstructions to those of several published state-of-the-art methods. We show that our method provides highly accurate reconstructions and identifies the number of clones, their genotypes and frequencies even at a modest sequencing depth. As a proof of concept, we apply our model to clinical data from three cases with chronic lymphocytic leukaemia and one case with acute myeloid leukaemia.

Motivation: In high-dimensional phenotyping screens, a large number of cellular features is observed after perturbing genes by knockouts or RNA interference. Comprehensive analysis of perturbation effects is one of the most powerful techniques for attributing functions to genes, but not much work has been done so far to adapt statistical and computational methodology to the specific needs of large-scale and high-dimensional phenotyping screens. Results: We introduce and compare probabilistic methods to efficiently infer a genetic hierarchy from the nested structure of observed perturbation effects. These hierarchies elucidate the structures of signaling pathways and regulatory networks. Our methods achieve two goals: (1) they reveal clusters of genes with highly similar phenotypic profiles, and (2) they order (clusters of) genes according to subset relationships between phenotypes. We evaluate our algorithms in the controlled setting of simulation studies and show their practical use in two experimental scenarios: (1) a data set investigating the response to microbial challenge in Drosophila melanogaster, and (2) a compendium of expression profiles of Saccharomyces cerevisiae knockout strains. We show that our methods identify biologically justified genetic hierarchies of perturbation effects. Availability: The software used in our analysis is freely available in the R package ‘nem’ from www.bioconductor.org Contact: ogt@cs.princeton.edu

How to link genotypes and phenotypes is a long-standing question in modern biology. Modern high-throughput approaches are key technologies at the forefront of genetic research. They enable the analysis of a biological response to thousands of experimental perturbations and require a tight collaboration between experimental and computational scientists. Perturbation studies and computational approaches have revolutionized research in functional genomics and genetics and promise to lay the foundation for personalized medicine. For modern high-throughput technologies, computation is as important as experimentation. Genome-wide image-based RNA interference (RNAi) screens, for example, are only feasible because of computational techniques. Computational skills to analyse the data have become as important as experimental skills to generate the data. Design and analysis of phenol typing screens depend on the number of genes perturbed and the richness of the phenotype observed (Figure 6.1). At one extreme are high-throughput screens with single reporters, e.g. a genome-wide screen for new components of a pathway. At the other extreme are perturbations of individual genes with very rich phenotypes, e.g. assessing the effects of a single gene perturbation on several molecular levels over time. Between these two extremes lie a variety of possible screen designs. Two widely used scenarios are small-scale perturbations ( Scenario 1: Genome-wide screens with single reporters RNAi screens have been frequently and successfully applied for functional profiling of genes on a large scale (Boutros & Ahringer 2008). The vast majority of these applications use a single phenotype (e.g. cell viability, growth rate, activity of reporter constructs) to characterize the function of genes in specific biological pathways.

A new study based on single-nucleus sequencing reports that triple-negative breast cancers acquire copy number aberrations in short punctuated bursts in the earliest stages of tumor evolution, rather than continuously and gradually, challenging prevailing models of tumor evolution.

Here, I argue that computational thinking and techniques are so central to the quest of understanding life that today all biology is computational biology. Computational biology brings order into our understanding of life, it makes biological concepts rigorous and testable, and it provides a reference map that holds together individual insights. The next modern synthesis in biology will be driven by mathematical, statistical, and computational methods being absorbed into mainstream biological training, turning biology into a quantitative science.

Cancer is a somatic evolutionary process characterized by the accumulation of mutations, which contribute to tumor growth, clinical progression, immune escape, and drug resistance development. Evolutionary theory can be used to analyze the dynamics of tumor cell populations and to make inference about the evolutionary history of a tumor from molecular data. We review recent approaches to modeling the evolution of cancer, including population dynamics models of tumor initiation and progression, phylogenetic methods to model the evolutionary relationship between tumor subclones, and probabilistic graphical models to describe dependencies among mutations. Evolutionary modeling helps to understand how tumors arise and will also play an increasingly important prognostic role in predicting disease progression and the outcome of medical interventions, such as targeted therapy.

Linking networks of molecular interactions to cellular functions and phenotypes is a key goal in systems biology. Here, we adapt concepts of spatial statistics to assess the functional content of molecular networks. Based on the guilt-by-association principle, our approach (called SANTA) quantifies the strength of association between a gene set and a network, and functionally annotates molecular networks like other enrichment methods annotate lists of genes. As a general association measure, SANTA can (i) functionally annotate experimentally derived networks using a collection of curated gene sets and (ii) annotate experimentally derived gene sets using a collection of curated networks, as well as (iii) prioritize genes for follow-up analyses. We exemplify the efficacy of SANTA in several case studies using the S. cerevisiae genetic interaction network and genome-wide RNAi screens in cancer cell lines. Our theory, simulations, and applications show that SANTA provides a principled statistical way to quantify the association between molecular networks and cellular functions and phenotypes. SANTA is available from http://bioconductor.org/packages/release/bioc/html/SANTA.html., Author Summary Molecular networks are maps of the tens of thousands of interactions that occur between the components of biological systems. Types of interactions include physical, genetic and functional interactions between genes, gene products and metabolites. Network-based approaches to molecular biology are increasingly being used to better understand cellular functions. Currently, gene set methods can be used to functionally annotate the hits from high-throughput studies; however, no methods exist to functionally annotate molecular interaction networks. This greatly limits our ability to quantify the often subtle functional adaptions that occur in networks as they rewire to respond to external stimuli. Here, we extend well-tested concepts from spatial statistics to define a general association measure between networks and gene sets. Like Gene Set Enrichment Analysis, our approach measures concordant changes, but does this on networks, rather than on lists of genes. We validate it both in simulations and real-world case studies. We apply our approach to genetic interaction networks mapped under different conditions and created using different methods, and demonstrate how it extends the previous analyses of data sets, allowing us to better understand the high-level changes that occur within cells.

Building better models of cellular pathways is one of the major challenges of systems biology and functional genomics. There is a need for methods to build on established expert knowledge and reconcile it with results of high-throughput studies. Moreover, the available data sources are heterogeneous and need to be combined in a way specific for the part of the pathway in which they are most informative. Here, we present a compartment specific strategy to integrate edge, node and path data for the refinement of a network hypothesis. Specifically, we use a local-move Gibbs sampler for refining pathway hypotheses from a compendium of heterogeneous data sources, including novel methodology for integrating protein attributes. We demonstrate the utility of this approach in a case study of the pheromone response MAPK pathway in the yeast S. cerevisiae.

Intra-tumour genetic heterogeneity is the result of ongoing evolutionary change within each cancer. The expansion of genetically distinct sub-clonal populations may explain the emergence of drug resistance, and if so, would have prognostic and predictive utility. However, methods for objectively quantifying tumour heterogeneity have been missing and are particularly difficult to establish in cancers where predominant copy number variation prevents accurate phylogenetic reconstruction owing to horizontal dependencies caused by long and cascading genomic rearrangements. To address these challenges, we present MEDICC, a method for phylogenetic reconstruction and heterogeneity quantification based on a Minimum Event Distance for Intra-tumour Copy-number Comparisons. Using a transducer-based pairwise comparison function, we determine optimal phasing of major and minor alleles, as well as evolutionary distances between samples, and are able to reconstruct ancestral genomes. Rigorous simulations and an extensive clinical study show the power of our method, which outperforms state-of-the-art competitors in reconstruction accuracy, and additionally allows unbiased numerical quantification of tumour heterogeneity. Accurate quantification and evolutionary inference are essential to understand the functional consequences of tumour heterogeneity. The MEDICC algorithms are independent of the experimental techniques used and are applicable to both next-generation sequencing and array CGH data., Author Summary Cancer is a disease of random mutation and selection within the cellular genomes of an organism. As a result, when advanced disease is diagnosed, the cells comprising the tumour show a great amount of variability on the genomic level, a phenomenon termed intra-tumour genetic heterogeneity. Heterogeneity is thought to be one of the main reasons why tumors become resistant to therapy, and thus hinders personalised medicine approaches. If we want to understand tumour heterogeneity and its connection to resistance development we need to quantify it, which implies reconstructing the evolutionary history of cancer within the patient. Unfortunately, so far, methods for accurate reconstructions of these particular evolutionary trees and for quantification of heterogeneity have been missing. We here present MEDICC, a method that uses a minimum evolution criterion to compare cancer genomes based on genomic profiles of DNA content (copy-number profiles). It enables accurate reconstruction of the history of the disease and quantifies heterogeneity. It is specifically designed to deal with diploid human genomes, in that it disentangles genomic events on both parental alleles and includes a variety of accompanying algorithms to test for shapes of the evolutionary trees as well as the rate at which the cancer evolves.

Cancer has long been understood as a somatic evolutionary process, but many details of tumor progression remain elusive. Here, I will talk about computational methods to infer evolutionary pathways leading to tumour heterogeneity. Using a study in ovarian cancer I will describe how summary statistics of phylogenetic quantification of genetic heterogeneity can predict patient survival. Then, I will present probabilistic frameworks to reconstruct intra-tumor evolution from data obtained by bulk bisulfite sequencing of mixed tumor samples or single cell genomes. Our approaches jointly estimates the number and composition of clones in the sample as well as the most likely tree connecting them. I will finish with an outlook how to put cancer genomes in a tissue context by using image analysis and deconvolution of gene expression data. Citation Format: Florian Markowetz. Quantifying tumor heterogeneity in the microenvironment. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr SY38-03.

Cancer has long been understood as a somatic evolutionary process, but many details of tumor progression remain elusive. Here, I will talk about computational methods to infer evolutionary pathways leading to tumor heterogeneity. I will start by discussing how to quantify intra-patient heterogeneity by rigorous analysis of copy-number profiles from multiple patient samples. Using a study in ovarian cancer I will describe how summary statistics of phylogenetic quantification of genetic heterogeneity can predict patient survival. Then, I will present a probabilistic framework to reconstruct intra-tumor evolution from data obtained by bulk bisulfite sequencing of mixed tumor samples or single cell genomes. Our approach jointly estimates the number and composition of clones in the sample as well as the most likely tree connecting them. I will finish by describing ongoing work to reconstruct patterns of clonal evolution from single cell sequencing data. Citation Format: Florian Markowetz. Quantifying patterns of tumor evolution. [abstract]. In: Proceedings of the AACR Special Conference on Tumor Metastasis; 2015 Nov 30-Dec 3; Austin, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(7 Suppl):Abstract nr IA20.

Solid tumors are heterogeneous tissues composed of a mixture of cancer and normal cells, which complicates the interpretation of their molecular profiles. Furthermore, tissue architecture is generally not reflected in molecular assays, rendering this rich information underused. To address these challenges, we developed a computational approach based on standard hematoxylin and eosin-stained tissue sections and demonstrated its power in a discovery and validation cohort of 323 and 241 breast tumors, respectively. To deconvolute cellular heterogeneity and detect subtle genomic aberrations, we introduced an algorithm based on tumor cellularity to increase the comparability of copy number profiles between samples. We next devised a predictor for survival in estrogen receptor-negative breast cancer that integrated both image-based and gene expression analyses and significantly outperformed classifiers that use single data types, such as microarray expression signatures. Image processing also allowed us to describe and validate an independent prognostic factor based on quantitative analysis of spatial patterns between stromal cells, which are not detectable by molecular assays. Our quantitative, image-based method could benefit any large-scale cancer study by refining and complementing molecular assays of tumor samples.

Motivation: Copy number alterations (CNAs) associated with cancer are known to contribute to genomic instability and gene deregulation. Integrating CNAs with gene expression helps to elucidate the mechanisms by which CNAs act and to identify the transcriptional downstream targets of CNAs. Such analyses can help to sort functional driver events from the many accompanying passenger alterations. However, the way CNAs affect gene expression can vary in different cellular contexts, for example between different subtypes of the same cancer. Thus, it is important to develop computational approaches capable of inferring differential connectivity of regulatory networks in different cellular contexts. Results: We propose a statistical deregulation model that integrates copy number and expression data of different disease subtypes to jointly model common and differential regulatory relationships. Our model not only identifies CNAs driving gene expression changes, but at the same time also predicts differences in regulation that distinguish one cancer subtype from the other. We implement our model in a penalized regression framework and demonstrate in a simulation study the feasibility and accuracy of our approach. Subsequently, we show that this model can identify both known and novel aspects of cross-talk between the ER and NOTCH pathways in ER-negative-specific deregulations, when compared with ER-positive breast cancer. This flexible model can be applied on other modalities such as methylation or microRNA and expression to disentangle cancer signaling pathways. Availability: The Bioconductor-compliant R package DANCE is available from www.markowetzlab.org/software/ Contact: yinyin.yuan@cancer.org.uk; florian.markowetz@cancer.org.uk

Background Genes might have different gene interactions in different cell conditions, which might be mapped into different networks. Differential analysis of gene networks allows spotting condition-specific interactions that, for instance, form disease networks if the conditions are a disease, such as cancer, and normal. This could potentially allow developing better and subtly targeted drugs to cure cancer. Differential network analysis with direct physical gene interactions needs to be explored in this endeavour. Results C3NET is a recently introduced information theory based gene network inference algorithm that infers direct physical gene interactions from expression data, which was shown to give consistently higher inference performances over various networks than its competitors. In this paper, we present, DC3net, an approach to employ C3NET in inferring disease networks. We apply DC3net on a synthetic and real prostate cancer datasets, which show promising results. With loose cutoffs, we predicted 18583 interactions from tumor and normal samples in total. Although there are no reference interactions databases for the specific conditions of our samples in the literature, we found verifications for 54 of our predicted direct physical interactions from only four of the biological interaction databases. As an example, we predicted that RAD50 with TRF2 have prostate cancer specific interaction that turned out to be having validation from the literature. It is known that RAD50 complex associates with TRF2 in the S phase of cell cycle, which suggests that this predicted interaction may promote telomere maintenance in tumor cells in order to allow tumor cells to divide indefinitely. Our enrichment analysis suggests that the identified tumor specific gene interactions may be potentially important in driving the growth in prostate cancer. Additionally, we found that the highest connected subnetwork of our predicted tumor specific network is enriched for all proliferation genes, which further suggests that the genes in this network may serve in the process of oncogenesis. Conclusions Our approach reveals disease specific interactions. It may help to make experimental follow-up studies more cost and time efficient by prioritizing disease relevant parts of the global gene network.

Background Deregulation between two different cell populations manifests itself in changing gene expression patterns and changing regulatory interactions. Accumulating knowledge about biological networks creates an opportunity to study these changes in their cellular context. Results We analyze re-wiring of regulatory networks based on cell population-specific perturbation data and knowledge about signaling pathways and their target genes. We quantify deregulation by merging regulatory signal from the two cell populations into one score. This joint approach, called JODA, proves advantageous over separate analysis of the cell populations and analysis without incorporation of knowledge. JODA is implemented and freely available in a Bioconductor package 'joda'. Conclusions Using JODA, we show wide-spread re-wiring of gene regulatory networks upon neocarzinostatin-induced DNA damage in Human cells. We recover 645 deregulated genes in thirteen functional clusters performing the rich program of response to damage. We find that the clusters contain many previously characterized neocarzinostatin target genes. We investigate connectivity between those genes, explaining their cooperation in performing the common functions. We review genes with the most extreme deregulation scores, reporting their involvement in response to DNA damage. Finally, we investigate the indirect impact of the ATM pathway on the deregulated genes, and build a hypothetical hierarchy of direct regulation. These results prove that JODA is a step forward to a systems level, mechanistic understanding of changes in gene regulation between different cell populations.

Different data types can offer complementary perspectives on the same biological phenomenon. In cancer studies, for example, data on copy number alterations indicate losses and amplifications of genomic regions in tumours, while transcriptomic data point to the impact of genomic and environmental events on the internal wiring of the cell. Fusing different data provides a more comprehensive model of the cancer cell than that offered by any single type. However, biological signals in different patients exhibit diverse degrees of concordance due to cancer heterogeneity and inherent noise in the measurements. This is a particularly important issue in cancer subtype discovery, where personalised strategies to guide therapy are of vital importance. We present a nonparametric Bayesian model for discovering prognostic cancer subtypes by integrating gene expression and copy number variation data. Our model is constructed from a hierarchy of Dirichlet Processes and addresses three key challenges in data fusion: (i) To separate concordant from discordant signals, (ii) to select informative features, (iii) to estimate the number of disease subtypes. Concordance of signals is assessed individually for each patient, giving us an additional level of insight into the underlying disease structure. We exemplify the power of our model in prostate cancer and breast cancer and show that it outperforms competing methods. In the prostate cancer data, we identify an entirely new subtype with extremely poor survival outcome and show how other analyses fail to detect it. In the breast cancer data, we find subtypes with superior prognostic value by using the concordant results. These discoveries were crucially dependent on our model's ability to distinguish concordant and discordant signals within each patient sample, and would otherwise have been missed. We therefore demonstrate the importance of taking a patient-specific approach, using highly-flexible nonparametric Bayesian methods., Author Summary The goal of personalised medicine is to develop accurate diagnostic tests that identify patients who can benefit from targeted therapies. To achieve this goal it is necessary to stratify cancer patients into homogeneous subtypes according to which molecular aberrations their tumours exhibit. Prominent approaches for subtype definition combine information from different molecular levels, for example data on DNA copy number changes with data on mRNA expression changes. This is called data fusion. We contribute to this field by proposing a unified model that fuses different data types, finds informative features and estimates the number of subtypes in the data. The main strength of our model comes from the fact that we assess for each patient whether the different data agree on a subtype or not. Competing methods combine the data without checking for concordance of signals. On a breast cancer and a prostate cancer data set we show that concordance of signals has strong influence on subtype definition and that our model allows to define prognostic subtypes that would have been missed otherwise.

Phylogenetic tree reconstruction is traditionally based on multiple sequence alignments (MSAs) and heavily depends on the validity of this information bottleneck. With increasing sequence divergence, the quality of MSAs decays quickly. Alignment-free methods, on the other hand, are based on abstract string comparisons and avoid potential alignment problems. However, in general they are not biologically motivated and ignore our knowledge about the evolution of sequences. Thus, it is still a major open question how to define an evolutionary distance metric between divergent sequences that makes use of indel information and known substitution models without the need for a multiple alignment. Here we propose a new evolutionary distance metric to close this gap. It uses finite-state transducers to create a biologically motivated similarity score which models substitutions and indels, and does not depend on a multiple sequence alignment. The sequence similarity score is defined in analogy to pairwise alignments and additionally has the positive semi-definite property. We describe its derivation and show in simulation studies and real-world examples that it is more accurate in reconstructing phylogenies than competing methods. The result is a new and accurate way of determining evolutionary distances in and beyond the twilight zone of sequence alignments that is suitable for large datasets., Comment: to appear in PLoS ONE

Embryonic stem cells (ESC) have the potential to self-renew indefinitely and to differentiate into any of the three germ layers. The molecular mechanisms for self-renewal, maintenance of pluripotency and lineage specification are poorly understood, but recent results point to a key role for epigenetic mechanisms. In this study, we focus on quantifying the impact of histone 3 acetylation (H3K9,14ac) on gene expression in murine embryonic stem cells. We analyze genome-wide histone acetylation patterns and gene expression profiles measured over the first five days of cell differentiation triggered by silencing Nanog, a key transcription factor in ESC regulation. We explore the temporal and spatial dynamics of histone acetylation data and its correlation with gene expression using supervised and unsupervised statistical models. On a genome-wide scale, changes in acetylation are significantly correlated to changes in mRNA expression and, surprisingly, this coherence increases over time. We quantify the predictive power of histone acetylation for gene expression changes in a balanced cross-validation procedure. In an in-depth study we focus on genes central to the regulatory network of Mouse ESC, including those identified in a recent genome-wide RNAi screen and in the PluriNet, a computationally derived stem cell signature. We find that compared to the rest of the genome, ESC-specific genes show significantly more acetylation signal and a much stronger decrease in acetylation over time, which is often not reflected in an concordant expression change. These results shed light on the complexity of the relationship between histone acetylation and gene expression and are a step forward to dissect the multilayer regulatory mechanisms that determine stem cell fate., Comment: accepted at PLoS Computational Biology

The influence of DNA cis-regulatory elements on a gene's expression has been intensively studied. However, little is known about expressions driven by trans-acting DNA hotspots. DNA hotspots harboring copy number aberrations are recognized to be important in cancer as they influence multiple genes on a global scale. The challenge in detecting trans-effects is mainly due to the computational difficulty in detecting weak and sparse trans-acting signals amidst co-occuring passenger events. We propose an integrative approach to learn a sparse interaction network of DNA copy-number regions with their downstream targets in a breast cancer dataset. Information from this network helps distinguish copy-number driven from copy-number independent expression changes on a global scale. Our result further delineates cis- and trans-effects in a breast cancer dataset, for which important oncogenes such as ESR1 and ERBB2 appear to be highly copy-number dependent. Further, our model is shown to be efficient and in terms of goodness of fit no worse than other state-of the art predictors and network reconstruction models using both simulated and real data., Comment: Accepted at IEEE International Conference on Bioinformatics & Biomedicine (BIBM 2010)

Gene expression profiling using micro-arrays is a modern approach for molecular diagnostics. In clinical micro-array studies, researchers aim to predict disease type, survival, or treatment response using gene expression profiles. In this process, they encounter a series of obstacles and pitfalls. This chapter reviews fundamental issues from machine learning and recommends a procedure for the computational aspects of a clinical micro-array study.

Gene expression profiling using micro-arrays is a modern approach for molecular diagnostics. In clinical micro-array studies, researchers aim to predict disease type, survival, or treatment response using gene expression profiles. In this process, they encounter a series of obstacles and pitfalls. This chapter reviews fundamental issues from machine learning and recommends a procedure for the computational aspects of a clinical micro-array study.

Motivation: High-throughput screens (HTS) by RNAi or small molecules are among the most promising tools in functional genomics. They enable researchers to observe detailed reactions to experimental perturbations on a genome-wide scale. While there is a core set of computational approaches used in many publications to analyze these data, a specialized software combining them and making them easily accessible has so far been missing. Results: Here we describe HTSanalyzeR, a flexible software to build integrated analysis pipelines for HTS data that contains over-representation analysis, gene set enrichment analysis, comparative gene set analysis and rich sub-network identification. HTSanalyzeR interfaces with commonly used pre-processing packages for HTS data and presents its results as HTML pages and network plots. Availability: Our software is written in the R language and freely available via the Bioconductor project at http://www.bioconductor.org. Contact: florian.markowetz@cancer.org.uk

Intra-tumor genetic heterogeneity is the result of ongoing evolutionary change within each cancer. The expansion of genetically distinct sub-clonal populations may explain the emergence of drug resistance and if so would have prognostic and predictive utility. However, methods for objectively quantifying tumor heterogeneity have been missing and are particularly difficult to establish in cancers where predominant copy number variation prevents accurate phylogenetic reconstruction because of horizontal dependencies caused by long and cascading genomic rearrangements. To address these challenges we have developed MEDICC, a method for phylogenetic reconstruction and heterogeneity quantification, which determines optimal phasing of major and minor alleles, computes evolutionary distances between samples, and reconstructs ancestral genomes. Rigorous simulations and an extensive clinical study show the power of our method, which outperforms state-of-the-art competitors in reconstruction accuracy and additionally allows unbiased numerical quantification of tumor heterogeneity. Using MEDICC we investigated the relationship between tumor heterogeneity and patient outcome in high-grade serous ovarian cancer. We found that tumor heterogeneity in this cancer is driven by ongoing clonal evolution with fully branched evolutionary trajectories that do not have clock-like evolutionary rates. Quantitative measures of clonal expansion and temporal heterogeneity were the strongest predictors of progression-free survival. We show in 2 patients that clonal expansion of a minor subclone that was present prior to chemotherapy led to clinical relapse. Citation Information: Mol Cancer Ther 2013;12(11 Suppl):PL05-03. Citation Format: Florian M. Markowetz. Phylogenetic quantification of intra-tumor heterogeneity predicts time to relapse in high-grade serous ovarian cancer. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2013 Oct 19-23; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2013;12(11 Suppl):Abstract nr PL05-03.

Modern high-throughput gene perturbation screens are key technologies at the forefront of genetic research. Combined with rich phenotypic descriptors they enable researchers to observe detailed cellular reactions to experimental perturbations on a genome-wide scale. This review surveys the current state-of-the-art in analyzing perturbation screens from a network point of view. We describe approaches to make the step from the parts list to the wiring diagram by using phenotypes for network inference and integrating them with complementary data sources. The first part of the review describes methods to analyze one- or low-dimensional phenotypes like viability or reporter activity; the second part concentrates on high-dimensional phenotypes showing global changes in cell morphology, transcriptome or proteome., Review based on ISMB 2009 tutorial; after two rounds of revision

In this article we highlight recent developments in computational functional genomics to identify networks of functionally related genes and proteins based on diverse sources of genomic data. Our specific focus is on statistical methods to identify genetic networks. We discuss integrated analysis of microarray datasets, methods to combine heterogeneous data sources, the analysis of high-dimensional phenotyping screens and describe efforts to establish a reliable and unbiased gold standard for method comparison and evaluation.

Visualization and analysis of molecular networks are both central to systems biology. However, there still exists a large technological gap between them, especially when assessing multiple network levels or hierarchies. Here we present RedeR, an R/Bioconductor package combined with a Java core engine for representing modular networks. The functionality of RedeR is demonstrated in two different scenarios: hierarchical and modular organization in gene co-expression networks and nested structures in time-course gene expression subnetworks. Our results demonstrate RedeR as a new framework to deal with the multiple network levels that are inherent to complex biological systems. RedeR is available from http://bioconductor.org/packages/release/bioc/html/RedeR.html.