SMBP Group Meeting: Steffen Schultze (Max Planck Institute)

Tuesday May 20, 2025, 2:00 PM → 3:00 PM America/New_York

3rd Floor Classroom (162 Fifth Avenue )

Topic: Bayesian electron density determination from single-molecule X-ray scattering images 

Abstract: Scattering experiments using ultrashort X-ray free electron laser (XFEL) pulses have opened a new path for the structure determination of a wide variety of specimens, including nano-crystals and entire viruses, approaching atomistic spatial and femtoseconds time resolution. However, random and unknown sample orientations as well as low signal to noise ratios have so far prevented a successful application to smaller specimens like single biomolecules. In fact, most approaches for structure determination from these experiments rely on determining the unknown orientations from the images alone, which becomes fundamentally impossible for the low photons counts obtained for small proteins (10-100 per image). We present a rigorous Bayesian approach to overcome these limitations, additionally taking into account intensity fluctuations, beam polarisation, irregular detector shapes, incoherent scattering, and background scattering. We demonstrate electron density determination of various model proteins at 2Å resolution from simulated scattering images. We also show that not only single structures but entire structural ensembles can be determined using our Bayesian approach. A further problem is posed by that fact that delivery of a specimen molecule to the micrometer sized beam focus is successful for typically less than 1% of the pulses, such that 99% of the images are “empty”, but nevertheless contain noise photons. Current analysis methods rely on selecting the hits from the remaining misses, which is prohibited for single biomolecules by the immense noise background. We demonstrate by applying our Bayesian approach that the need for hit selection is circumvented, by marginalising over the possible presence or absence of the specimen in the likelihood function.