Seminar

High Performance Computing methods in cryo Electron Tomography

Daniel Castaño Díez, PhD

Cryo ET allows the visualization of biomolecules in a close-to-native state in their cellular environment. Besides adequate sample preparation and imaging techniques, the success of cryo-ET projects is highly dependent on the numerical methods used for the analysis of the raw data. The task of processing large amounts of threedimensional data is very demanding computationally. This requires both the mathematical development of performance-oriented algorithms and their adaption to the computer architecture requirements of high performace computation resources, as large scale parallel clusters of graphic accelerators (GPUS). In this talk we discuss the development of high-performing analysis tools specifically tailored for different stages of the processing pipeline in tomography: alignment of tilt series without fiducial markers (Allignator package), averaging of subtomograms (Dynamo package) and fitting of X-ray structures into low-resolution density maps.