Statistical Methods for Analysis of Array CGH Data

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Project Summary: The primary goal of this proposal is to develop computationally tractable methods of statistical analysis for array comparative genomic hybridization (aCGH) data at both the individual level of "signal processing" and the population level of detecting patterns. At the individual level of analysis, we aim to improve upon currently available methods through a simultaneous analysis of multiple chromosomes and hybridizations that exploits features that are shared in common, while accounting for variability within and between chromosomes and between hybridizations. At the population-level, we will develop novel methods for locating common regions of genomic instability and for clustering patients using clinical endpoints, such as survival. These methods are motivated by, and will be applied to, aCGH data sets from glioma studies and meningioma studies. Relevance: Malignant gliomas are the most common primary human brain tumors. Problems in their pathological classification, however, complicate patient management and have sparked considerable interest in molecular diagnostic approaches. Our group is currently developing methods for aCGH that, we hypothesize, can provide a sensitive, specific, cost-effective and rapid method to assess human malignant gliomas for relevant genetic changes. Meningioma, a common intracranial tumor found frequently in patients with neurofibromatosis type 2 (NF2), also occurs sporadically in individuals without germline NF2 mutations. It is necessary to seek genetic mechanisms that may operate in the initiation and progression of these sporadic meningiomas. In addition, aCGH profiling will likely be useful for differential diagnosis of familial multiple meningioma. Array CGH holds promise for uncovering small imbalanced chromosomal events in tumors and can provide specific information about the boundaries of the imbalanced chromosome segments (ICS). Sound statistical methods are required for efficient and valid analyses of these important data.