Model-based clustering of array CGH data

Variational Inference for Coupled Hidden Markov Models Applied to the Joint Detection of Copy Number Variations

Hidden Markov models provide a natural statistical framework for the detection of the copy number variations (CNV) in genomics. In this context, we define a hidden Markov process that underlies all individuals jointly in order to detect and to classify genomics regions in different states (typically, deletion, normal or amplification). Structural variations from different individuals may be dependent. It is the case in agronomy where varietal selection program exists and species share a common phylogenetic past. We propose to take into account these dependencies inthe HMM model. When dealing with a large number of series, maximum likelihood inference (performed classically using the EM algorithm) becomes intractable. We thus propose an approximate inference algorithm based on a variational approach (VEM), implemented in the CHMM R package. A simulation study is performed to assess the performance of the proposed method and an application to the detection of structural variations in plant genomes is presented.

Novel multisample scheme for inferring phylogenetic markers from whole genome tumor profiles

Computational cancer phylogenetics seeks to enumerate the temporal sequences of aberrations in tumor evolution, thereby delineating the evolution of possible tumor progression pathways, molecular subtypes, and mechanisms of action. We previously developed a pipeline for constructing phylogenies describing evolution between major recurring cell types computationally inferred from whole-genome tumor profiles. The accuracy and detail of the phylogenies, however, depend on the identification of accurate, high-resolution molecular markers of progression, i.e., reproducible regions of aberration that robustly differentiate different subtypes and stages of progression. Here, we present a novel hidden Markov model (HMM) scheme for the problem of inferring such phylogenetically significant markers through joint segmentation and calling of multisample tumor data. Our method classifies sets of genome-wide DNA copy number measurements into a partitioning of samples into normal (diploid) or amplified at each probe. It differs from other similar HMM methods in its design specifically for the needs of tumor phylogenetics, by seeking to identify robust markers of progression conserved across a set of copy number profiles. We show an analysis of our method in comparison to other methods on both synthetic and real tumor data, which confirms its effectiveness for tumor phylogeny inference and suggests avenues for future advances.

Parallel evolution of tumour subclones mimics diversity between tumours

Intratumour heterogeneity (ITH) may foster tumour adaptation and compromise the efficacy of personalized medicine approaches. The scale of heterogeneity within a tumour (intratumour heterogeneity) relative to genetic differences between tumours (intertumour heterogeneity) is unknown. To address this, we obtained 48 biopsies from eight stage III and IV clear cell renal cell carcinomas (ccRCCs) and used DNA copy-number analyses to compare biopsies from the same tumour with 440 single tumour biopsies from the Cancer Genome Atlas (TCGA). Unsupervised hierarchical clustering of TCGA and multi-region ccRCC samples revealed segregation of samples from the same tumour into unrelated clusters; 25% of multi-region samples appeared more similar to unrelated samples than to any other sample originating from the same tumour. We found that the majority of recurrent DNA copy number driver aberrations in single biopsies were not present ubiquitously in late-stage ccRCCs and were likely to represent subclonal events acquired during tumour progression. Such heterogeneous subclonal genetic alterations within individual tumours may impair the identification of robust ccRCC molecular subtypes classified by distinct copy number alterations and clinical outcomes. The co-existence of distinct subclonal copy number events in different regions of individual tumours reflects the diversification of individual ccRCCs through multiple evolutionary routes and may contribute to tumour sampling bias and impact upon tumour progression and clinical outcome.

A hidden Markov model-based algorithm for identifying tumour subtype using array CGH data

Background

The recent advancement in array CGH (aCGH) research has significantly improved tumor identification using DNA copy number data. A number of unsupervised learning methods have been proposed for clustering aCGH samples. Two of the major challenges for developing aCGH sample clustering are the high spatial correlation between aCGH markers and the low computing efficiency. A mixture hidden Markov model based algorithm was developed to address these two challenges.

Results

The hidden Markov model (HMM) was used to model the spatial correlation between aCGH markers. A fast clustering algorithm was implemented and real data analysis on glioma aCGH data has shown that it converges to the optimal cluster rapidly and the computation time is proportional to the sample size. Simulation results showed that this HMM based clustering (HMMC) method has a substantially lower error rate than NMF clustering. The HMMC results for glioma data were significantly associated with clinical outcomes.

Conclusions

We have developed a fast clustering algorithm to identify tumor subtypes based on DNA copy number aberrations. The performance of the proposed HMMC method has been evaluated using both simulated and real aCGH data. The software for HMMC in both R and C++ is available in ND INBRE website http://ndinbre.org/programs/bioinformatics.php.

Joint segmentation, calling, and normalization of multiple CGH profiles

The statistical analysis of array comparative genomic hybridization (CGH) data has now shifted to the joint assessment of copy number variations at the cohort level. Considering multiple profiles gives the opportunity to correct for systematic biases observed on single profiles, such as probe GC content or the so-called "wave effect." In this article, we extend the segmentation model developed in the univariate case to the joint analysis of multiple CGH profiles. Our contribution is multiple: we propose an integrated model to perform joint segmentation, normalization, and calling for multiple array CGH profiles. This model shows great flexibility, especially in the modeling of the wave effect that gives a likelihood framework to approaches proposed by others. We propose a new dynamic programming algorithm for break point positioning, as well as a model selection criterion based on a modified bayesian information criterion proposed in the univariate case. The performance of our method is assessed using simulated and real data sets. Our method is implemented in the R package cghseg.

Preprocessing and downstream analysis of microarray DNA copy number profiles

Analysis of DNA copy number profiles requires methods tailored to the specific nature of these data. The number of available data analysis methods has grown enormously in the last 5 years. We discuss the typical characteristics of DNA copy number data, as measured by microarray technology and review the extensive literature on preprocessing methods such as segmentation and calling. Subsequently, the focus narrows to applications of DNA copy number in cancer, in particular, several downstream analyses of multi-sample data sets such as testing, clustering and classification. Finally, we look ahead: what should we prepare for and which methodology-related topics may deserve attention in the near future?

Genome-wide analysis of cutaneous T-cell lymphomas identifies three clinically relevant classes

This study was undertaken to identify recurrent genetic alterations of the three main types of cutaneous T-cell lymphomas (CTCLs): mycosis fungoides (MF), Sézary syndrome (SS), and cutaneous anaplastic large-cell lymphoma (CALCL). Using array-based comparative genomic hybridization, the molecular cytogenetic profiles of 72 samples obtained from 58 patients with CTCL corresponding to 24 transformed MF (T-MF), 16 SS, and 18 CALCLs were determined. T-MF was characterized by gains of 1q25-31, 7p22-11.2, 7q21, 7q31, and 17q12, and losses of 9p21, 10p11.2, and 10q26. SS exhibited gains of 8q23-24.3 and 17q23-24, as well as losses of 9p21, 10p12-11.2, 10q22-24, 10q25-26, and 17p13-q11.1. Finally, CALCL exhibited 6q27 and 13q34 losses. Such imbalances were statistically associated with one CTCL subtype. Unsupervised hierarchical clustering defined three categories of clinical relevance: (1) CALCL apart from epidermotropic-CTCL, (2) an SS-only category, and (3) a mixed category with T-MF and SS cases, with both primary and secondary SS cases. In rare cases, the genetic classification did not correspond to the inclusion diagnosis, possibly reflecting the association of two diseases in the same patient or initial misdiagnosis according to follow-up. Finally, different samples in the same patient clustered together, showing reproducibility of such a classifier.