A stationary wavelet entropy-based clustering approach accurately predicts gene expression

TY - JOUR

T1 - A stationary wavelet entropy-based clustering approach accurately predicts gene expression

AU - Nguyen, Nha

AU - Vo, An

AU - Choi, Inchan

AU - Won, Kyoung-Jae

PY - 2015/3/1

Y1 - 2015/3/1

N2 - Studying epigenetic landscapes is important to understand the condition for gene regulation. Clustering is a useful approach to study epigenetic landscapes by grouping genes based on their epigenetic conditions. However, classical clustering approaches that often use a representative value of the signals in a fixed-sized window do not fully use the information written in the epigenetic landscapes. Clustering approaches to maximize the information of the epigenetic signals are necessary for better understanding gene regulatory environments. For effective clustering of multidimensional epigenetic signals, we developed a method called Dewer, which uses the entropy of stationary wavelet of epigenetic signals inside enriched regions for gene clustering. Interestingly, the gene expression levels were highly correlated with the entropy levels of epigenetic signals. Dewer separates genes better than a window-based approach in the assessment using gene expression and achieved a correlation coefficient above 0.9 without using any training procedure. Our results show that the changes of the epigenetic signals are useful to study gene regulation.

AB - Studying epigenetic landscapes is important to understand the condition for gene regulation. Clustering is a useful approach to study epigenetic landscapes by grouping genes based on their epigenetic conditions. However, classical clustering approaches that often use a representative value of the signals in a fixed-sized window do not fully use the information written in the epigenetic landscapes. Clustering approaches to maximize the information of the epigenetic signals are necessary for better understanding gene regulatory environments. For effective clustering of multidimensional epigenetic signals, we developed a method called Dewer, which uses the entropy of stationary wavelet of epigenetic signals inside enriched regions for gene clustering. Interestingly, the gene expression levels were highly correlated with the entropy levels of epigenetic signals. Dewer separates genes better than a window-based approach in the assessment using gene expression and achieved a correlation coefficient above 0.9 without using any training procedure. Our results show that the changes of the epigenetic signals are useful to study gene regulation.

KW - 3T3-L1 Cells

KW - Algorithms

KW - Animals

KW - Cluster Analysis

KW - Entropy

KW - Epigenesis, Genetic

KW - Gene Expression

KW - Mice

KW - Models, Genetic

KW - Multigene Family

KW - Promoter Regions, Genetic

KW - Wavelet Analysis

U2 - 10.1089/cmb.2014.0221

DO - 10.1089/cmb.2014.0221

M3 - Journal article

C2 - 25383910

SN - 1066-5277

VL - 22

SP - 236

EP - 249

JO - Journal of computational biology : a journal of computational molecular cell biology

JF - Journal of computational biology : a journal of computational molecular cell biology

IS - 3

ER -