An accurate and reliable method for automatic picking of HIV/SIV spikes

Abstract

The HIV/SIV envelope spikes initiate infection in the host cells and hence understanding their structure is one of the primary focus areas of modern research in structural biology. The structures of the native envelope spikes are still unknown. A mutant form of SIV virions has 80 to 90 envelope spikes per virion; HIV virions possess only 8–9 spikes. Thus, any electron tomographic study of the spike structure becomes very tedious if spikes are selected by hand for further processing. Consequently automating the process of spike selection is very important for determining the spike structure in situ. To this end, we have developed a nearly automated procedure for spike selection based on a “segmentation by classification” philosophy.