Avocado: a multi-scale deep tensor factorization method learns a latent representation of the human epigenome

摘要

The Human Genome Project, at its completion in 2003, yielded an accurate description of the nucleotide sequence of the human genome but an incomplete picture of how that sequence operates within each cell. Characterizing each basepair of the genome with just two bits of information—its nucleotide identity—yielded many critical insights into genome biology but also left open a host of questions about how this static view of the genome gives rise to a diversity of cell types. Clearly, answering these questions required gathering more data.