Details

“Single Cells, Many Paths: The Role of Data Science in Drug Discovery”

We face a challenging reality that it takes $3.2 billion and 12 years on average to develop a new medicine that is approved for sale in the United States. Computational research holds promise to improve our ability to make medicines – through predictions of toxicity, generative design of molecules, mining public research, among many other examples. In this presentation I’ll focus on data science in the very first step of drug discovery: identifying disease-associated genes to drug. Leveraging genome-wide data from single cells, we use stochastic optimization and deep neural networks to create cell atlases. These cell atlases can be used to identify novel cell types, infer disease-associated cell state changes, and predict the effects of gene perturbations via transfer learning. We applied these techniques to identify anti-inflammatory genes for progressive multiple sclerosis, and further characterized the disease with dynamic spatial imaging from mouse tissue sections. I’ll conclude with comments on data science in the biotech and pharmaceutical industry – what careers look like, how to train toward this field, and the great opportunities for computational scientists to reimagine the paths towards innovative medicines for patients.

Speaker Bio

Sarah Spencer received a B.A. in biology from Washington University in St. Louis in 2009, and then pursued a PhD from MIT in computational and systems biology from 2011–2017. For the next three years, she joined CRISPR Therapeutics, researching gene editing and supporting the first clinical trials using CRISPR technology. She spent a year at Alltrna, a start-up engineering RNA drugs, prior to joining Novartis in 2022. As an associate director and senior principal scientist at Novartis, Sarah leads a computational research team focused on the very earliest stages of drug discovery.