Explore cutting-edge approaches to predicting single-cell responses to perturbations in this comprehensive lecture and primer from the Broad Institute. Delve into the world of neural optimal transport methods for modeling cellular changes, presented by Charlotte Bunne from ETH Zurich. Learn how these advanced deep learning techniques achieve state-of-the-art results in predicting treatment responses at the single-cell level, with applications in large-scale clinical studies. Discover the analytical challenges and opportunities in studying cell state transitions using single-cell genomics, presented by Oana Ursu from Genentech. Gain insights into key questions in the field, including quantifying perturbation effects, predicting combinatorial perturbations, and modeling cell population responses. Understand how computational advances synergize with experimental techniques to provide high-resolution insights into cell states across multiple modalities, time, and space.
Neural Optimal Transport for Inferring Single-Cell Responses to Perturbations
Overview
Syllabus
Introduction
How do cells change between different states
What determines cell transitions
Identifying regulators of cell transitions
Experimental methods
Single cell genomics
Types of perturbations
Abstract cell state space
Linear regression
Intuition
Nonlinearity
Perturbation Myth
Errors
Connection to networks
Parallel efforts
Gene expression programs
Major pitfalls
Overfitting
Cell Types
Validation
Predictability
Transfer Learning
genomoid screens
Neural optimal transport
Taught by
Broad Institute
