Dataset Information

Robust unmixing of tumor states in array comparative genomic hybridization data.

ABSTRACT:

Motivation

Tumorigenesis is an evolutionary process by which tumor cells acquire sequences of mutations leading to increased growth, invasiveness and eventually metastasis. It is hoped that by identifying the common patterns of mutations underlying major cancer sub-types, we can better understand the molecular basis of tumor development and identify new diagnostics and therapeutic targets. This goal has motivated several attempts to apply evolutionary tree reconstruction methods to assays of tumor state. Inference of tumor evolution is in principle aided by the fact that tumors are heterogeneous, retaining remnant populations of different stages along their development along with contaminating healthy cell populations. In practice, though, this heterogeneity complicates interpretation of tumor data because distinct cell types are conflated by common methods for assaying the tumor state. We previously proposed a method to computationally infer cell populations from measures of tumor-wide gene expression through a geometric interpretation of mixture type separation, but this approach deals poorly with noisy and outlier data.

Results

In the present work, we propose a new method to perform tumor mixture separation efficiently and robustly to an experimental error. The method builds on the prior geometric approach but uses a novel objective function allowing for robust fits that greatly reduces the sensitivity to noise and outliers. We further develop an efficient gradient optimization method to optimize this 'soft geometric unmixing' objective for measurements of tumor DNA copy numbers assessed by array comparative genomic hybridization (aCGH) data. We show, on a combination of semi-synthetic and real data, that the method yields fast and accurate separation of tumor states.

Conclusions

We have shown a novel objective function and optimization method for the robust separation of tumor sub-types from aCGH data and have shown that the method provides fast, accurate reconstruction of tumor states from mixed samples. Better solutions to this problem can be expected to improve our ability to accurately identify genetic abnormalities in primary tumor samples and to infer patterns of tumor evolution.

Supplementary information

Supplementary data are available at Bioinformatics online.

SUBMITTER: Tolliver D

PROVIDER: S-EPMC2881397 | biostudies-literature | 2010 Jun

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Robust unmixing of tumor states in array comparative genomic hybridization data.

Tolliver David D Tsourakakis Charalampos C Subramanian Ayshwarya A Shackney Stanley S Schwartz Russell R

Bioinformatics (Oxford, England) 20100601 12

<h4>Motivation</h4>Tumorigenesis is an evolutionary process by which tumor cells acquire sequences of mutations leading to increased growth, invasiveness and eventually metastasis. It is hoped that by identifying the common patterns of mutations underlying major cancer sub-types, we can better understand the molecular basis of tumor development and identify new diagnostics and therapeutic targets. This goal has motivated several attempts to apply evolutionary tree reconstruction methods to assay ...[more]

PMID: 20529894

Similar Datasets

Project description:Array Comparative Genomic Hybridization (aCGH) is a rapid screening technique to detect gene deletions and duplications, providing an overview of chromosomal aberrations throughout the entire genome of a tumor, without the need for cell culturing. However, the heterogeneity of aCGH data obfuscates existing methods of data analysis. Analysis of aCGH data from a systems biology perspective or in the context of total aberrations is largely absent in the published literature. We present here a novel alternative to the functional analysis of aCGH data using the phylogenetic paradigm that is well-suited to high dimensional datasets of heterogeneous nature, but has not been widely adapted to aCGH data. Maximum parsimony phylogenetic analysis sorts out genetic data through the simplest presentation of the data on a cladogram, a graphical evolutionary tree, thus providing a powerful and efficient method for aCGH data analysis. For example, the cladogram models the multiphasic changes in the cancer genome and identifies shared early mutations in the disease progression, providing a simple yet powerful means of aCGH data interpretation. As such, applying maximum parsimony phylogenetic analysis to aCGH results allows for the differentiation between drivers and passenger genes aberrations in cancer specimens. In addition to offering a novel methodology to analyze aCGH results, we present here a crucial software suite that we wrote to carry out the analysis. In a broader context, we wish to underscore that phylogenetic analysis of aCGH data is a non-parametric method that circumvents the pitfalls and frustrations of standard analytical techniques that rely on parametric statistics. Organizing the data in a cladogram as explained in this research article provides insights into the disease common aberrations, as well as the disease subtypes and their shared aberrations (the synapomorphies) of each subtype. Hence, we report the method and make the software suite publicly and freely available at http://software.phylomcs.com so that researchers can test alternative and innovative approaches to the analysis of aCGH data.

Project description:Breast cancer is a profoundly heterogeneous disease with respect to biological and clinical behavior. Gene expression profiling has been used to dissect this complexity and stratify tumors into intrinsic gene expression subtypes associated with distinct biology, patient outcome and different genomic alterations. Additionally, breast tumors occurring in individuals with germline BRCA1 or BRCA2 mutations typically fall into distinct subtypes. We applied global DNA copy number and gene expression profiling in 359 breast tumors. All tumors were classified according to intrinsic gene expression subtypes and included cases from genetically predisposed women. The Genomic Identification of Significant Targets in Cancer (GISTIC) algorithm was used to identify significant DNA copy number aberrations and genomic subgroups of breast cancer. We identified 31 genomic regions that were highly amplified in >1% of the 359 breast tumors. Several amplicons were found to co-occur, the 8p12 and 11q13.3 regions being the most frequent combination besides amplicons on the same chromosomal arm. Unsupervised hierarchical clustering with 133 significant GISTIC regions (66 and 67 with DNA copy number gain and loss, respectively) revealed six genomic subtypes, termed: 17q12, basal-complex, luminal-simple, luminal-complex, amplifier and mixed subtype. Four of them had striking similarity to intrinsic gene expression subtypes and showed association to conventional tumor biomarkers and clinical outcome. However, luminal A-classified tumors were distributed in two main genomic subtypes, luminal-simple and luminal-complex, the former group having better prognosis while the latter group included also luminal B and the majority of BRCA2-mutated tumors. The basal-complex subtype displayed extensive genomic homogeneity and harbored the majority of BRCA1-mutated tumors. The 17q12 subtype comprised mostly HER2-amplified and HER2-enriched subtype tumors and had the worst prognosis. The amplifier and mixed subtypes contained tumors from all gene expression subtypes, the former being enriched for 8p12-amplified cases while the mixed subtype included many tumors with predominantly DNA copy number losses and poor prognosis. Genomic profiling of 359 breast tumors using tiling BAC aCGH. A number of cases were hybridized as replicates or replicate as dye-swaps. Gene expression profiling of 359 breast tumors using 55K oligonucleotide microarrays.

Project description:We designed a study to investigate genetic relationships between primary tumors of oral squamous cell carcinoma (OSCC) and their lymph node metastases, and to identify genomic copy number aberrations (CNAs) related to lymph node metastasis. For this purpose, we collected a total of 42 tumor samples from 25 patients and analyzed their genomic profiles by array-based comparative genomic hybridization. We then compared the genetic profiles of metastatic primary tumors (MPTs) with their paired lymph node metastases (LNMs), and also those of LNMs with non-metastatic primary tumors (NMPTs). Firstly, we found that although there were some distinctive differences in the patterns of genomic profiles between MPTs and their paired LNMs, the paired samples shared similar genomic aberration patterns in each case. Unsupervised hierarchical clustering analysis grouped together 12 of the 15 MPT-LNM pairs. Furthermore, similarity scores between paired samples were significantly higher than those between non-paired samples. These results suggested that MPTs and their paired LNMs are composed predominantly of genetically clonal tumor cells, while minor populations with different CNAs may also exist in metastatic OSCCs. Secondly, to identify CNAs related to lymph node metastasis, we compared CNAs between grouped samples of MPTs and LNMs, but were unable to find any CNAs that were more common in LNMs. Finally, we hypothesized that subpopulations carrying metastasis-related CNAs might be present in both the MPT and LNM. Accordingly, we compared CNAs between NMPTs and LNMs, and found that gains of 7p, 8q and 17q were more common in the latter than in the former, suggesting that these CNAs may be involved in lymph node metastasis of OSCC. In conclusion, our data suggest that in OSCCs showing metastasis, the primary and metastatic tumors share similar genomic profiles, and that cells in the primary tumor may tend to metastasize after acquiring metastasis-associated CNAs.

Dataset Information

Robust unmixing of tumor states in array comparative genomic hybridization data.

Motivation

Results

Conclusions

Supplementary information

Publications

Robust unmixing of tumor states in array comparative genomic hybridization data.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets