Avantika Lal, NVIDIA
gtc-dc 2019
We’ll discuss the application of deep learning to genomic datasets, a rapidly-growing practice primed to revolutionize genomic analysis. The NVIDIA genomics team is building deep learning tools that are improving, accelerating, and reducing the cost of genomic analysis across sequencing instruments and applications. We’ll present two examples of our research, the first being a deep neural network for variant calling, which matches the performance of state-of-the-art variant calling tools. The second is a deep learning model that denoises low-quality ATAC-Seq data, reducing the cost of ATAC-Seq and enabling high-quality sequencing results from small numbers of cells.