Jiarui Ding

The clonal and mutational evolution spectrum of primary triple-negative breast cancers.

Primary triple-negative breast cancers (TNBCs), a tumour type defined by lack of oestrogen receptor, progesterone receptor and ERBB2 gene amplification, represent approximately 16% of all breast cancers. Here we show in 104 TNBC cases that at the time of diagnosis these cancers exhibit a wide and continuous spectrum of genomic evolution, with some having only a handful of coding somatic aberrations in a few pathways, whereas others contain hundreds of coding somatic mutations. High-throughput RNA sequencing (RNA-seq) revealed that only approximately 36% of mutations are expressed. Using deep re-sequencing measurements of allelic abundance for 2,414 somatic mutations, we determine for the first time—to our knowledge—in an epithelial tumour subtype, the relative abundance of clonal frequencies among cases representative of the population. We show that TNBCs vary widely in their clonal frequencies at the time of diagnosis, with the basal subtype of TNBC showing more variation than non-basal TNBC. Although p53 (also known as TP53), PIK3CA and PTEN somatic mutations seem to be clonally dominant compared to other genes, in some tumours their clonal frequencies are incompatible with founder status. Mutations in cytoskeletal, cell shape and motility proteins occurred at lower clonal frequencies, suggesting that they occurred later during tumour progression. Taken together, our results show that understanding the biology and therapeutic responses of patients with TNBC will require the determination of individual tumour clonal genotypes.

Distinct evolutionary trajectories of primary high-grade serous ovarian cancers revealed through spatial mutational profiling

High‐grade serous ovarian cancer (HGSC) is characterized by poor outcome, often attributed to the emergence of treatment‐resistant subclones. We sought to measure the degree of genomic diversity within primary, untreated HGSCs to examine the natural state of tumour evolution prior to therapy. We performed exome sequencing, copy number analysis, targeted amplicon deep sequencing and gene expression profiling on 31 spatially and temporally separated HGSC tumour specimens (six patients), including ovarian masses, distant metastases and fallopian tube lesions. We found widespread intratumoural variation in mutation, copy number and gene expression profiles, with key driver alterations in genes present in only a subset of samples (eg PIK3CA, CTNNB1, NF1). On average, only 51.5% of mutations were present in every sample of a given case (range 10.2–91.4%), with TP53 as the only somatic mutation consistently present in all samples. Complex segmental aneuploidies, such as whole‐genome doubling, were present in a subset of samples from the same individual, with divergent copy number changes segregating independently of point mutation acquisition. Reconstruction of evolutionary histories showed one patient with mixed HGSC and endometrioid histology, with common aetiologic origin in the fallopian tube and subsequent selection of different driver mutations in the histologically distinct samples. In this patient, we observed mixed cell populations in the early fallopian tube lesion, indicating that diversity arises at early stages of tumourigenesis. Our results revealed that HGSCs exhibit highly individual evolutionary trajectories and diverse genomic tapestries prior to therapy, exposing an essential biological characteristic to inform future design of personalized therapeutic solutions and investigation of drug‐resistance mechanisms.

Mutational and structural analysis of diffuse large B-cell lymphoma using whole-genome sequencing

Diffuse large B-cell lymphoma (DLBCL) is a genetically heterogeneous cancer composed of at least 2 molecular subtypes that differ in gene expression and distribution of mutations. Recently, application of genome/exome sequencing and RNA-seq to DLBCL has revealed numerous genes that are recurrent targets of somatic point mutation in this disease. Here we provide a whole-genome-sequencing-based perspective of DLBCL mutational complexity by characterizing 40 de novo DLBCL cases and 13 DLBCL cell lines and combining these data with DNA copy number analysis and RNA-seq from an extended cohort of 96 cases. Our analysis identified widespread genomic rearrangements including evidence for chromothripsis as well as the presence of known and novel fusion transcripts. We uncovered new gene targets of recurrent somatic point mutations and genes that are targeted by focal somatic deletions in this disease. We highlight the recurrence of germinal center B-cell-restricted mutations affecting genes that encode the S1P receptor and 2 small GTPases (GNA13 and GNAI2) that together converge on regulation of B-cell homing. We further analyzed our data to approximate the relative temporal order in which some recurrent mutations were acquired and demonstrate that ongoing acquisition of mutations and intratumoral clonal heterogeneity are common features of DLBCL. This study further improves our understanding of the processes and pathways involved in lymphomagenesis, and some of the pathways mutated here may indicate new avenues for therapeutic intervention.

Use of mutation profiles to refine the classification of endometrial carcinomas.

The classification of endometrial carcinomas is based on pathological assessment of tumour cell type; the different cell types (endometrioid, serous, carcinosarcoma, mixed, undifferentiated, and clear cell) are associated with distinct molecular alterations. This current classification system for high‐grade subtypes, in particular the distinction between high‐grade endometrioid (EEC‐3) and serous carcinomas (ESC), is limited in its reproducibility and prognostic abilities. Therefore, a search for specific molecular classifiers to improve endometrial carcinoma subclassification is warranted. We performed target enrichment sequencing on 393 endometrial carcinomas from two large cohorts, sequencing exons from the following nine genes: ARID1A, PPP2R1A, PTEN, PIK3CA, KRAS, CTNNB1, TP53, BRAF, and PPP2R5C. Based on this gene panel, each endometrial carcinoma subtype shows a distinct mutation profile. EEC‐3s have significantly different frequencies of PTEN and TP53 mutations when compared to low‐grade endometrioid carcinomas. ESCs and EEC‐3s are distinct subtypes with significantly different frequencies of mutations in PTEN, ARID1A, PPP2R1A, TP53, and CTNNB1. From the mutation profiles, we were able to identify subtype outliers, ie cases diagnosed morphologically as one subtype but with a mutation profile suggestive of a different subtype. Careful review of these diagnostically challenging cases suggested that the original morphological classification was incorrect in most instances. The molecular profile of carcinosarcomas suggests two distinct mutation profiles for these tumours: endometrioid‐type (PTEN, PIK3CA, ARID1A, KRAS mutations) and serous‐type (TP53 and PPP2R1A mutations). While this nine‐gene panel does not allow for a purely molecularly based classification of endometrial carcinoma, it may prove useful as an adjunct to morphological classification and serve as an aid in the classification of problematic cases. If used in practice, it may lead to improved diagnostic reproducibility and may also serve to stratify patients for targeted therapeutics. Copyright

JointSNVMix: a probabilistic model for accurate detection of somatic mutations in normal/tumour paired next-generation sequencing data.

Motivation: Identification of somatic single nucleotide variants (SNVs) in tumour genomes is a necessary step in defining the mutational landscapes of cancers. Experimental designs for genome-wide ascertainment of somatic mutations now routinely include next-generation sequencing (NGS) of tumour DNA and matched constitutional DNA from the same individual. This allows investigators to control for germline polymorphisms and distinguish somatic mutations that are unique to the tumour, thus reducing the burden of labour-intensive and expensive downstream experiments needed to verify initial predictions. In order to make full use of such paired datasets, computational tools for simultaneous analysis of tumour–normal paired sequence data are required, but are currently under-developed and under-represented in the bioinformatics literature. Results: In this contribution, we introduce two novel probabilistic graphical models called JointSNVMix1 and JointSNVMix2 for jointly analysing paired tumour–normal digital allelic count data from NGS experiments. In contrast to independent analysis of the tumour and normal data, our method allows statistical strength to be borrowed across the samples and therefore amplifies the statistical power to identify and distinguish both germline and somatic events in a unified probabilistic framework. Availability: The JointSNVMix models and four other models discussed in the article are part of the JointSNVMix software package available for download at http://compbio.bccrc.ca Contact: sshah@bccrc.ca Supplementary information:Supplementary data are available at Bioinformatics online.

Integrative analysis of genome-wide loss of heterozygosity and monoallelic expression at nucleotide resolution reveals disrupted pathways in triple-negative breast cancer

Loss of heterozygosity (LOH) and copy number alteration (CNA) feature prominently in the somatic genomic landscape of tumors. As such, karyotypic aberrations in cancer genomes have been studied extensively to discover novel oncogenes and tumor-suppressor genes. Advances in sequencing technology have enabled the cost-effective detection of tumor genome and transcriptome mutation events at single-base-pair resolution; however, computational methods for predicting segmental regions of LOH in this context are not yet fully explored. Consequently, whole transcriptome, nucleotide-level resolution analysis of monoallelic expression patterns associated with LOH has not yet been undertaken in cancer. We developed a novel approach for inference of LOH from paired tumor/normal sequence data and applied it to a cohort of 23 triple-negative breast cancer (TNBC) genomes. Following extensive benchmarking experiments, we describe the nucleotide-resolution landscape of LOH in TNBC and assess the consequent effect of LOH on the transcriptomes of these tumors using RNA-seq-derived measurements of allele-specific expression. We show that the majority of monoallelic expression in the transcriptomes of triple-negative breast cancer can be explained by genomic regions of LOH and establish an upper bound for monoallelic expression that may be explained by other tumor-specific modifications such as epigenetics or mutations. Monoallelically expressed genes associated with LOH reveal that cell cycle, homologous recombination and actin-cytoskeletal functions are putatively disrupted by LOH in TNBC. Finally, we show how inference of LOH can be used to interpret allele frequencies of somatic mutations and postulate on temporal ordering of mutations in the evolutionary history of these tumors.

TITAN: inference of copy number architectures in clonal cell populations from tumor whole-genome sequence data

The evolution of cancer genomes within a single tumor creates mixed cell populations with divergent somatic mutational landscapes. Inference of tumor subpopulations has been disproportionately focused on the assessment of somatic point mutations, whereas computational methods targeting evolutionary dynamics of copy number alterations (CNA) and loss of heterozygosity (LOH) in whole-genome sequencing data remain underdeveloped. We present a novel probabilistic model, TITAN, to infer CNA and LOH events while accounting for mixtures of cell populations, thereby estimating the proportion of cells harboring each event. We evaluate TITAN on idealized mixtures, simulating clonal populations from whole-genome sequences taken from genomically heterogeneous ovarian tumor sites collected from the same patient. In addition, we show in 23 whole genomes of breast tumors that the inference of CNA and LOH using TITAN critically informs population structure and the nature of the evolving cancer genome. Finally, we experimentally validated subclonal predictions using fluorescence in situ hybridization (FISH) and single-cell sequencing from an ovarian cancer patient sample, thereby recapitulating the key modeling assumptions of TITAN.

Ovarian and endometrial endometrioid carcinomas have distinct CTNNB1 and PTEN mutation profiles

Ovarian endometrioid carcinomas and endometrial endometrioid carcinomas share many histological and molecular alterations. These similarities are likely due to a common endometrial epithelial precursor cell of origin, with most ovarian endometrioid carcinomas arising from endometriosis. To directly compare the mutation profiles of two morphologically similar tumor types, endometrial endometrioid carcinomas (n=307) and ovarian endometrioid carcinomas (n=33), we performed select exon capture sequencing on a panel of genes: ARID1A, PTEN, PIK3CA, KRAS, CTNNB1, PPP2R1A, TP53. We found that PTEN mutations are more frequent in low-grade endometrial endometrioid carcinomas (67%) compared with low-grade ovarian endometrioid carcinomas (17%) (P<0.0001). By contrast, CTNNB1 mutations are significantly different in low-grade ovarian endometrioid carcinomas (53%) compared with low-grade endometrial endometrioid carcinomas (28%) (P<0.0057). This difference in CTNNB1 mutation frequency may be reflective of the distinct microenvironments; the epithelial cells lining an endometriotic cyst within the ovary are exposed to a highly oxidative environment that promotes tumorigenesis. Understanding the distinct mutation patterns found in the PI3K and Wnt pathways of ovarian and endometrial endometrioid carcinomas may provide future opportunities for stratifying patients for targeted therapeutics.

DriverNet: uncovering the impact of somatic driver mutations on transcriptional networks in cancer

Simultaneous interrogation of tumor genomes and transcriptomes is underway in unprecedented global efforts. Yet, despite the essential need to separate driver mutations modulating gene expression networks from transcriptionally inert passenger mutations, robust computational methods to ascertain the impact of individual mutations on transcriptional networks are underdeveloped. We introduce a novel computational framework, DriverNet, to identify likely driver mutations by virtue of their effect on mRNA expression networks. Application to four cancer datasets reveals the prevalence of rare candidate driver mutations associated with disrupted transcriptional networks and a simultaneous modulation of oncogenic and metabolic networks, induced by copy number co-modification of adjacent oncogenic and metabolic drivers. DriverNet is available on Bioconductor or at http://compbio.bccrc.ca/software/drivernet/.

Feature-based classifiers for somatic mutation detection in tumour–normal paired sequencing data

Motivation: The study of cancer genomes now routinely involves using next-generation sequencing technology (NGS) to profile tumours for single nucleotide variant (SNV) somatic mutations. However, surprisingly few published bioinformatics methods exist for the specific purpose of identifying somatic mutations from NGS data and existing tools are often inaccurate, yielding intolerably high false prediction rates. As such, the computational problem of accurately inferring somatic mutations from paired tumour/normal NGS data remains an unsolved challenge. Results: We present the comparison of four standard supervised machine learning algorithms for the purpose of somatic SNV prediction in tumour/normal NGS experiments. To evaluate these approaches (random forest, Bayesian additive regression tree, support vector machine and logistic regression), we constructed 106 features representing 3369 candidate somatic SNVs from 48 breast cancer genomes, originally predicted with naive methods and subsequently revalidated to establish ground truth labels. We trained the classifiers on this data (consisting of 1015 true somatic mutations and 2354 non-somatic mutation positions) and conducted a rigorous evaluation of these methods using a cross-validation framework and hold-out test NGS data from both exome capture and whole genome shotgun platforms. All learning algorithms employing predictive discriminative approaches with feature selection improved the predictive accuracy over standard approaches by statistically significant margins. In addition, using unsupervised clustering of the ground truth ‘false positive’ predictions, we noted several distinct classes and present evidence suggesting non-overlapping sources of technical artefacts illuminating important directions for future study. Availability: Software called MutationSeq and datasets are available from http://compbio.bccrc.ca. Contact: saparicio@bccrc.ca Supplementary information: Supplementary data are available at Bioinformatics online.