Deep Learning and Accelerated Computing for Single-Cell Genomic Data
, NVIDIA
Understanding the role of regulatory DNA in disease is critical for drug discovery. Epigenomic methods such as ATAC sequencing, which measures DNA accessibility, allow us to identify regulatory DNA elements that control genes. In recent years, single-cell ATAC sequencing studies have revealed disease mechanisms in specific cell types within the human body. However, even as single-cell datasets grow, current methods are slow to scale and perform poorly at understanding rare cell types. I'll discuss NVIDIA’s effort to apply deep learning and GPU acceleration to large-scale epigenomic data analysis. I'll show how NVIDIA's RAPIDS libraries can be used to accelerate discovery of cell types in the human body. In addition, we developed AtacWorks, a deep learning toolkit to enhance ATAC-seq data and identify active regulatory DNA more accurately than existing state-of-the-art methods. These methods can be combined to create powerful, interactive tools for epigenetic data analysis.