Manifold-embedding methods for extracting continuous conformational ensembles of biological molecules from single-particle measurements using X-ray Free Electron Lasers.

A novel machine-learning approach allows us to navigate the high-dimensional space of single-particle XFEL scattering data. This technique can be used to map continuous conformational changes in biological systems, and to determine the energy landscape associated with such changes. With the extremely large datasets expected from high repetition-rate XFELs about to enter service, this approach promises unprecedented access to rare, rate-limiting conformations energetically far above the thermal bath.