Hi,
The particle sets we are working with have an intrinsically low signal-to-noise ratio. While we were able to achieve reasonable 2D classifications, the results of ab initio reconstructions were inconsistent across different settings. Are there any effective strategies for generating ab initio reconstructions using particle sets with low SNR? Additionally, how can we validate that an ab initio reconstruction accurately represents the true particle architecture?
Based on my experience, the correct ab initio model often had a larger number of particles compared to other suggested models (out of 3 or 5), and subsequent homogeneous refinement quickly improved the resolution, yielding a reasonable density map.
Any suggestions would be appreciated.
Uhn-Soo
