Machine-learning model maps human gene initiator sequences

UC San Diego researchers used sequencing data from roughly 500,000 initiator variants to train a machine-learning model that predicts the DNA pattern of the initiator region. They report the sequence in about 60% of human genes. The authors say the data and model can help predict mutation effects tied to the initiator and design customized synthetic promoters for future laboratory experiments. The study was published in Genes & Development.