FSC-work/FSC-free gap, and curves not converging near FSC=0.143

Troubleshooting
1

image
image661×330 29.3 KB

I have two related questions about this FSC plot (helical fibrils, ~2.6–2.7 Å final resolution, Phenix real-space refine cross-validation, RELION 5):

  1. There is a big, systematic gap between FSC-work and FSC-free — confirmed from the raw per-shell data (not just the plot), roughly 0.13–0.18 in the 3–5 Å and 2–2.5 Å bins.

  2. Near FSC = 0.143, the four curves aren’t closely converged — FSC-free in particular drops noticeably faster than FSC-masked/FSC-sum through the mid-resolution range, widening the spread by the time it reaches the 0.143 threshold.

My suspicion is that these two observations are actually the same underlying problem viewed two ways — the free curve dropping unusually fast in the mid-resolution range is what’s producing both the work/free gap and the uneven spread near 0.143 — rather than two separate issues.

Thanks in advance for any input!