Automatic identification of informative regions with epigenomic changes associated to hematopoiesis

Hematopoiesis is one of the best characterized biological systems but the connection between chromatin changes and lineage differentiation is not yet well understood. We have developed a bioinformatic workflow to generate a chromatin space that allows to classify forty-two human healthy blood epigenomes from the BLUEPRINT, NIH ROADMAP and ENCODE consortia by their cell type. This approach recapitulates the human hematopoietic differentiation tree model from an epigenomic perspective. The analysis of the orthogonal dimension of the chromatin space allows us to identify 32,662 chromatin determinant regions (CDRs), genomic regions with different epigenetic characteristics between the cell types. Functional analysis revealed that these regions are linked with cell identities. The inclusion of leukemia epigenomes in the healthy hematological chromatin sample space gives us insights on the origin of these tumors. Our method provides an analytical approach to study the relationship between epigenomic changes and cell lineage differentiation. Method availability: https://github.com/david-juan/ChromDet