The Incredible Convergence Of Deep Learning And Genomics

In 2014, few of us worked at the intersection of deep learning and genomics. Three years later, genomics is in the midst of a paradigm shift -- deep learning for genomics is coming. How did we get here? In late 2014, we developed our first working deep learning for genomics model -- the "Chromputer". Chromputer used CNNs similar to AlexNet to predict histone modifications and chromatin states from 2D DNA accessibility data (ATAC-seq).