MC-Bayes: A Python-based wrapper for MotionCor3 processing of EER files compatible with Bayesian polishing

MC-Bayes is a Python-based script designed to streamline the processing of cryo-electron microscopy EER movies on one or multiple GPUs using MotionCor3. This innovative tool generates the essential .star files required for RELION to perform Bayesian polishing, also known as reference-based motion correction, on EER movies. Prior to the development of MC-Bayes, the only option for Bayesian polishing of EER data entailed using the CPU-based "RELIONCor" implementation of MotionCor2, which proved to be suboptimal on GPU-heavy cryo-EM processing systems.

The primary motivation behind the creation of MC-Bayes was to address the needs of facilities and users who possess a significant number of powerful GPUs but have limited CPU cores or insufficient system RAM. By harnessing the capabilities of MotionCor3, MC-Bayes enables motion correction of EER data at a rate of two or more times faster than the RELION CPU implementation, under most circumstances.

This improvement becomes even more pronounced when dealing with systems equipped with dozens or hundreds of CPU cores and substantial quantities of system RAM, making the GPU-based approach the preferred choice for efficient processing in cryo-EM laboratories.