Helex: an Evolutionary Tabu Search Strategy for the Identification of Helical Regions in cryo-Electron Microscopy Reconstructions

Abstract

Cryo-electron microscopy (cryo-EM) enables the imaging of macromolecular complexes in near-native environments at resolutions that in many instances approach atomic level of detail. Already covering 9.8% of the EBI database for cryo-EM reconstructions, the maps at intermediate- to high-resolution (better than 8A) potentially allow the identification of secondary structure elements. Especially, alpha helices which frequently show consistent patterns in volumetric maps, may be annotated using pattern matching methods. Previously introduced approaches predict secondary structure elements in cryo-EM datasets by applying multi-step heuristics to characterize the geometric features of the volumetric data. Here, we introduce helex (Helix Extractor) - a novel technique for the identification of helical regions in cryo-EM data sets. Helex is a hybrid optimization technique that combines a genetic algorithm, a tabu-search strategy, and a one-dimensional iterative optimization to locate and characterize helical regions. Our method takes advantage of the stochastic nature of genetic algorithms to identify optimal placements for a template helix. These placements are then used to characterize the length of the helical region, using an adaptive one-dimensional search that allows suboptimal steps during the optimization. Moreover, the tabu-search strategy prevents further exploration of already characterized helical regions. The method was extensively evaluated using maps at various resolutions, sampling rates and system complexity. Helex reliably identified medium to large helices (> 6 amino acids) in both synthetic and experimental test cases, placing them frequently with root mean square deviations of 1A or less from the correct helical axes. Helex is distributed in our molecular modeling software Sculptor, freely available at http://sculptor.biomachina.org.