Minimum error calibration and normalization for genomic copy number analysis

Modeling metastatic progression from cross-sectional cancer genomics data

MOTIVATION

Metastasis formation is a hallmark of cancer lethality. Yet, metastases are generally unobservable during their early stages of dissemination and spread to distant organs. Genomic datasets of matched primary tumors and metastases may offer insights into the underpinnings and the dynamics of metastasis formation.

RESULTS

We present metMHN, a cancer progression model designed to deduce the joint progression of primary tumors and metastases using cross-sectional cancer genomics data. The model elucidates the statistical dependencies among genomic events, the formation of metastasis, and the clinical emergence of both primary tumors and their metastatic counterparts. metMHN enables the chronological reconstruction of mutational sequences and facilitates estimation of the timing of metastatic seeding. In a study of nearly 5000 lung adenocarcinomas, metMHN pinpointed TP53 and EGFR as mediators of metastasis formation. Furthermore, the study revealed that post-seeding adaptation is predominantly influenced by frequent copy number alterations.

AVAILABILITY AND IMPLEMENTATION

All datasets and code are available on GitHub at https://github.com/cbg-ethz/metMHN.

Candidate targets of copy number deletion events across 17 cancer types

Genome variation is the direct cause of cancer and driver of its clonal evolution. While the impact of many point mutations can be evaluated through their modification of individual genomic elements, even a single copy number aberration (CNA) may encompass hundreds of genes and therefore pose challenges to untangle potentially complex functional effects. However, consistent, recurring and disease-specific patterns in the genome-wide CNA landscape imply that particular CNA may promote cancer-type-specific characteristics. Discerning essential cancer-promoting alterations from the inherent co-dependency in CNA would improve the understanding of mechanisms of CNA and provide new insights into cancer biology and potential therapeutic targets. Here we implement a model using segmental breakpoints to discover non-random gene coverage by copy number deletion (CND). With a diverse set of cancer types from multiple resources, this model identified common and cancer-type-specific oncogenes and tumor suppressor genes as well as cancer-promoting functional pathways. Confirmed by differential expression analysis of data from corresponding cancer types, the results show that for most cancer types, despite dissimilarity of their CND landscapes, similar canonical pathways are affected. In 25 analyses of 17 cancer types, we have identified 19 to 169 significant genes by copy deletion, including RB1, PTEN and CDKN2A as the most significantly deleted genes among all cancer types. We have also shown a shared dependence on core pathways for cancer progression in different cancers as well as cancer type separation by genome-wide significance scores. While this work provides a reference for gene specific significance in many cancers, it chiefly contributes a general framework to derive genome-wide significance and molecular insights in CND profiles with a potential for the analysis of rare cancer types as well as non-coding regions.

BoBafit: a copy number clustering tool designed to refit and recalibrate the baseline region of tumors’ profiles

•

Tools generating CN profiles derive the baseline region from samples’ median CN signal.

•

This canonical approach might cause wrongly estimated CN profile in complex karyotypes.

•

CNAs call is crucial for patients’ risk stratification aimed at personalized treatment.

•

BoBafit computes the correct baseline region and the CN profile, taking into account tumor genomic complexity and samples-specific alterations.

•

BoBafit should be implemented within CN analysis pipelines especially for clinical aims.

Human cancer arises from a population of cells that have acquired a wide range of genetic alterations, most of which are targets of therapeutic treatments or are used as prognostic factors for patient’s risk stratification. Among these, copy number alterations (CNAs) are quite frequent. Currently, several molecular biology technologies, such as microarrays, NGS and single-cell approaches are used to define the genomic profile of tumor samples. Output data need to be analyzed with bioinformatic approaches and particularly by employing computational algorithms.

Molecular biology tools estimate the baseline region by comparing either the mean probe signals, or the number of reads to the reference genome. However, when tumors display complex karyotypes, this type of approach could fail the baseline region estimation and consequently cause errors in the CNAs call. To overcome this issue, we designed an R-package, BoBafit, able to check and, eventually, to adjust the baseline region, according to both the tumor-specific alterations’ context and the sample-specific clustered genomic lesions.

Several databases have been chosen to set up and validate the designed package, thus demonstrating the potential of BoBafit to adjust copy number (CN) data from different tumors and analysis techniques.

Relevantly, the analysis highlighted that up to 25% of samples need a baseline region adjustment and a redefinition of CNAs calls, thus causing a change in the prognostic risk classification of the patients.

We support the implementation of BoBafit within CN analysis bioinformatics pipelines to ensure a correct patient’s stratification in risk categories, regardless of the tumor type.

Signatures of Discriminative Copy Number Aberrations in 31 Cancer Subtypes

Copy number aberrations (CNA) are one of the most important classes of genomic mutations related to oncogenetic effects. In the past three decades, a vast amount of CNA data has been generated by molecular-cytogenetic and genome sequencing based methods. While this data has been instrumental in the identification of cancer-related genes and promoted research into the relation between CNA and histo-pathologically defined cancer types, the heterogeneity of source data and derived CNV profiles pose great challenges for data integration and comparative analysis. Furthermore, a majority of existing studies have been focused on the association of CNA to pre-selected "driver" genes with limited application to rare drivers and other genomic elements. In this study, we developed a bioinformatics pipeline to integrate a collection of 44,988 high-quality CNA profiles of high diversity. Using a hybrid model of neural networks and attention algorithm, we generated the CNA signatures of 31 cancer subtypes, depicting the uniqueness of their respective CNA landscapes. Finally, we constructed a multi-label classifier to identify the cancer type and the organ of origin from copy number profiling data. The investigation of the signatures suggested common patterns, not only of physiologically related cancer types but also of clinico-pathologically distant cancer types such as different cancers originating from the neural crest. Further experiments of classification models confirmed the effectiveness of the signatures in distinguishing different cancer types and demonstrated their potential in tumor classification.