SFSC (Spectral SSNR-based Fourier Shell Correlation) based alignment and clustering of cryo-electron subtomograms

Abstract

The quality of a 3-D structure obtained by averaging a subtomogram cluster in Cryo-Electron Tomography majorly depends upon the alignment of the subtomograms within the cluster and contamination concerning the membership of particles with similar structural organization. SFSC is a robust and fast quality score that can efficiently rank subtomogram clusters of large sizes. In a comparative analysis of more than 15 scoring functions, it was found that SFSC was better and more efficient in ranking subtomogram clusters when compared to other scoring functions frequently used as a standard to assess the quality of a cluster. Currently, constrained cross correlation is used as a standard alignment metric to align subtomograms in a cluster. However, it has limitations like statistical fluctuations in considering missing wedges due to the random splitting of the set of subtomograms into two and the ability to work on limited sample size due to its pairwise alignment nature. These limitations can be tackled by using SFSC as a metric to align the subtomograms in a cluster to produce high-quality scores indicating better quality of the subtomogram averages. In this study, we intend to develop an alignment algorithm based on the SFSC function, implement it on simulated and experimental cellular subtomogram clusters, and compare it to the existing constrained cross-correlation method (missing wedge as a constraint), a standard method used for subtomogram alignment, in terms of the resolution of subtomogram averages produced by the two methods.