Fig. 4

COCOA can be applied to multi-omics analyses that include epigenetic data. a COCOA can annotate latent factors that were not annotated by a gene set approach. In the top of panel a, dark blue indicates that the data type explained at least 1% of the variation of the latent factor while light blue indicates that the data type explained between 0.1 and 1% of the variation. Gray indicates less than 0.1% explained. In the bottom of panel a, green indicates that at least one statistically significant gene set or region set was found for the latent factor and gray indicates no significant gene or region sets were found. b COCOA identifies an enhancer region set from a transformed B lymphocyte cell line where DNA methylation is correlated with latent factor 1 and IGHV mutation status, a marker of mature B cells that have undergone somatic hypermutation. The 50 CpGs with the highest absolute correlation with LF1 from the region set are shown. c Meta-region profiles show covariation between DNA methylation and LF8 score in certain regions bound by transcription factors functional in stem cell biology and by H3K4me1 in a stem cell line compared to the surrounding genome. The number of regions from each region set that were covered by epigenetic data in the COCOA analysis is indicated by “n”