Add to My Interests
Remove from My Interests
We'll explain how single cell genomics is revolutionizing the study of molecular biology and identification of treatments for disease, with a focus on how we use deep learning to generate novel insight in these studies. We trained associative domain adaptation neural networks on single cell RNA genomics datasets to predict how malaria cells will respond to different perturbations in order to identify treatments that will stop its transmission. We'll discuss how we trained variational autoencoders and BiGANs to build quantitative models of the cell to predict the most efficient way to turn stem-like cells into different cell types, and we'll provide an example of reprogramming cancer-like cells back to normal. Given the orders of magnitude of single cell genomic data being generated compared to previous technologies, GPU acceleration is critical to performing these tasks.
Do Not Sell My Personal Information