31.312.

.28. The decision tree of the random forest model for the oil content data.

enetic-genetic interplay pattern discovery

were downloaded from GEO. They are GSE36001 [Kresse, et al.,

SE53261 [Wagner, et al., 2014] and GSE68909. GSE36001

human osteosarcoma cell lines. There were 48,701 genes and

methylation sites. The study identified differentially expressed

d differentially methylated CpG sites. In this chapter, only genes

body methylation sites and genes with the genomic base-pair

information available were used for the demonstration of the

pigenetic interplay pattern discovery. This led to 23,215 genes

hylation sites for the analysis. The data were composed of 19

oma cell lines and six normal cell lines. Because samples were

d, the cross-replicate method was used to calculate differential

ns for each gene and to calculate differential methylations for

hylation site. For instance, the 19 osteosarcoma cell lines were

d with each of six normal cell lines to calculate 19 differential

n or methylation for one gene or one methylation site. Finally a

al expression matrix of 114 columns and a differential

on matrix of 114 columns were generated for identifying the

ally expressed genes and the differentially methylated sites.

ifferentially expressed genes (DEGs) were identified using the

ackage with a critical fold change value one and a critical p value