Overview

Part of the Image-Based Scientific Machine Learning for Theories of Biological Dynamics Across Scales program

Many parts of the modern cellular imaging toolkit are only available in genetically editable and lab-culturable cell lines and organisms. However, these cell lines and organisms make up only the tiniest fraction of the diversity of life on Earth and are often adapted to life in the laboratory. How do we learn about the physics and biodynamics of life at the cellular scale from non-model organisms using label-free imaging modalities such as brightfield, darkfield, DIC, phase contrast, and electron microscopy that lack molecular specificity but are still rich sources of biological dynamics? Are there ways to pair cellular-scale imaging data with genetic and phylogenetic data to infer what proteins impart different functions? What behaviors are conserved across species? What approaches that we have previously taken can be generalized and updated? What could we learn about the physical principles governing behaviors with help from machine learning?

This workshop will bring together experimentalists and theorists working in these systems.

The workshop will have two key goals:

Develop community standards and methodological priorities for creating accessible, standardized, machine-learning-ready datasets of cellular dynamics across diverse non-model organisms; identify where new machine-learning and computational approaches are needed; and communicate these needs and best practices to the broader community.
Catalyze new collaborations by forming small interdisciplinary groups of experimentalists, computational researchers, and mathematicians to tackle specific challenges identified during the workshop.

Image-Based Scientific Machine Learning for Theories of Biological Dynamics Across Scales – Equation discovery from images of non-model organisms

Overview

Attendees

Organizers

Participants