Skip to main content
Fig. 3 | BMC Bioinformatics

Fig. 3

From: A convolutional neural network-based regression model to infer the epigenetic crosstalk responsible for CG methylation patterns

Fig. 3

Cross-species application of epiNet models to human oocytes. a Pearson correlation coefficients between the predicted and actual CG methylation patterns of human FGOs. The features used to predict the CG methylation pattern were H3K4me3, H3K27me3 and chromatin accessibility. The species from which the training and test data originated are indicated. When training and testing were performed in the same species, the test data were from chromosome 1 and training and validation data were from the rest of the chromosomes. For cross-species testing, H3K4me3, H3K27me3 and chromatin accessibility data from all mouse chromosomes other than chromosome 1 were used to build an epiNet model. Then, H3K4me3, H3K27me3 and chromatin accessibility data from the entire human genome were used to predict the human CG methylation pattern. b A representative genome browser shot showing predicted CG methylation patterns of human FGOs. The actual CG methylation, H3K4me3 enrichment, H3K27me3 enrichment, and chromatin accessibility of human FGOs are shown for comparison. RefSeq genes are indicated at the bottom

Back to article page