“Classification and Alignment of Sub-Tomograms”
Achilleas Frangakis – Goethe University, Frankfurt Institute of Biophysics
lassification of electron sub-tomograms is a challenging task, because of the missing information and the low signal-to-noise ratio. Classification algorithms tend to classify data according to their orientation to the missing-wedge, rather than to the underlying signal. In this talk I will present a neural network approach and compare it with PCA classification, which is used to classify sub-tomograms. The performance of this algorithm will be discussed as well as various pitfalls involved in the alignment of the sub-tomograms. Finally I will show how classification algorithms can be used to cross-validate the results of template-matching by classifying sub-tomograms extracted from cellular tomograms of Mycoplasma pneumoniae.