Dataset Information

Incorporating higher-order representative features improves prediction in network-based cancer prognosis analysis.

ABSTRACT:

Background

In cancer prognosis studies with gene expression measurements, an important goal is to construct gene signatures with predictive power. In this study, we describe the coordination among genes using the weighted coexpression network, where nodes represent genes and nodes are connected if the corresponding genes have similar expression patterns across samples. There are subsets of nodes, called modules, that are tightly connected to each other. In several published studies, it has been suggested that the first principal components of individual modules, also referred to as "eigengenes", may sufficiently represent the corresponding modules.

Results

In this article, we refer to principal components and their functions as representative features". We investigate higher-order representative features, which include the principal components other than the first ones and second order terms (quadratics and interactions). Two gradient thresholding methods are adopted for regularized estimation and feature selection. Analysis of six prognosis studies on lymphoma and breast cancer shows that incorporating higher-order representative features improves prediction performance over using eigengenes only. Simulation study further shows that prediction performance can be less satisfactory if the representative feature set is not properly chosen.

Conclusions

This study introduces multiple ways of defining the representative features and effective thresholding regularized estimation approaches. It provides convincing evidence that the higher-order representative features may have important implications for the prediction of cancer prognosis.

SUBMITTER: Ma S

PROVIDER: S-EPMC3037289 | biostudies-literature | 2011 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Incorporating higher-order representative features improves prediction in network-based cancer prognosis analysis.

Ma Shuangge S Kosorok Michael R MR Huang Jian J Dai Ying Y

BMC medical genomics 20110112

<h4>Background</h4>In cancer prognosis studies with gene expression measurements, an important goal is to construct gene signatures with predictive power. In this study, we describe the coordination among genes using the weighted coexpression network, where nodes represent genes and nodes are connected if the corresponding genes have similar expression patterns across samples. There are subsets of nodes, called modules, that are tightly connected to each other. In several published studies, it h ...[more]

PMID: 21226928

Similar Datasets

Project description:It is increasingly being recognized that physical interactions between distant chromosomal loci influence the expressivity of the genome. Thus, multiple enhancers may converge on a single gene promoter 1,2 and a single enhancer can stochastically loop to multiple promoters 3. Regulatory elements from neighboring domains or from other chromosomes may interact to generate chromatin structures poised for transcription4-8 or repression 9-11. However, a broader perspective on the formation and function of such networks remains largely unexplored. Using the high-resolution 4C (Circular Chromosome Conformation Capture) analysis, we identify here a genome-wide physical network between genomically imprinted domains impinging H19 imprinting control region (ICR). Sequences interacting with the H19 ICR constitutively map at or within most of known imprinted domains both in mouse embryonic stem cells and in vitro-derived embryoid bodies, despite distinct differences in chromatin environments. Moreover, all-to-all 3D DNA FISH analyses revealed that imprinted domains from 7 different chromosomes co-localize with each other in pair-wise patterns, highlighting the dynamic nature of these interactions. The observed interaction network is dependent on CTCF binding sites within the maternally inherited H19 ICR allele and delays reprogramming of epigenetic features at imprinted domains during male germline development. As other imprinted regions can interact with each other in an H19 ICR-dependent manner, the H19 ICR emerges as an organizer of chromosomal networks by specifically recognizing and reconfiguring imprinted domains. We propose that an ancient key regulatory region, such as the H19 ICR, gradually impacted epigenetic modifications via long-range interactions in the germline to facilitate the recruitment of imprinted genes and their formation in clusters during mammalian radiation. Furthermore, the perspective of epigenetic lesions de-stabilizing such chromosomal networks may advance our understanding of how quantitative traits influence human diseases, such as cancer.

Dataset Information

Incorporating higher-order representative features improves prediction in network-based cancer prognosis analysis.

Background

Results

Conclusions

Publications

Incorporating higher-order representative features improves prediction in network-based cancer prognosis analysis.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets