Figure 1

Overview of the data and analyses. (A) I collected existing genome-wide maps of two histone marks, H3K4me1 and H3K27ac, from stages of a directed differentiation of mouse embryonic stem (ES) cells into cardiomyocytes, from heart tissues collected from several life stages, and from several other tissues. I evaluated how well these marks, which are associated with enhancer activity, could predict experimentally validated heart enhancers in E11.5 mice (“Target”). (B) I took a supervised machine learning approach to this problem by constructing feature vectors for validated enhancers and control regions based on the presence or absence of these histone modifications at their genomic locations. I created classifiers based on different subsets of the data from the cellular contexts given in (A) and evaluated them using cross validation.