Deep learning approaches are proving to be an important tool for tackling problems in genomics. Specifically, predicting fundamental molecular phenomena, such as gene transcription, or epigenetic marks) from DNA sequence is an important problem in understanding genome function. We will describe DeepMind’s approach to leveraging large amounts of epigenetic information to predict gene expression from DNA sequence.
DeepMind is committed to tackling important unsolved problems in biology, and in science more broadly. We will describe a variety of deep learning models developed by DeepMind (such as AlphaFold), which have solved important outstanding scientific problems. We will also discuss the impact these models have had in real-world applications.