How does MC find reference points?


I am trying to reprocess EMPIAR 10872. After importing, and just observing the micrographs I cannot see any particles present (which doesnt mean they arent there).

However, after going through motion correction and CTF estimation, there appears to be no particles found. Specifically, the MC shows motion on the +/-1 pixel and huddled around the center showing no significant motion at all.

After CTF estimation, no defocus optimization is found.

Assuming there are in fact particles present, it seems that MC is not finding the particles in which to track.

  1. How does MC find and track reference points?
  2. Is there a way to tune this so it can find these points?

Any insight would be appreciated,


Have you assessed the dataset in its entirety? Is this representative of the dataset?

Judging from the manuscript, 2885 movies were used, far fewer than the 3800+ in the EMPIAR deposition. Often, a proportion of micrographs are deemed suboptimal due to empty/thick/crystalline ice, limited information content, proximity to focus, drift, etc, and discarded during preprocessing. Is it possible this could be one of the suboptimal ones?

If all the micrographs appear like this, it may be worth double-checking the parameters at import.



Absolutely check other micrographs. Chances are you are looking at one which just doesn’t have anything on it… either a hole with no ice was imaged (in which case it is motion correcting detector noise), or the ice was extremely thin and vaporised as soon as the beam hit it (same thing) or it’s really, really close to focus.

I don’t think the last one is the case, however, as I’ve had micrographs extremely close to focus and they’ve estimated correctly.

Looking at the provided images my initial impression is that that micrograph is just detector noise.

Hi @mjmcleod64,

Motion correction in CryoSPARC estimates alignments by computing the correlation between movie frames; it doesn’t attempt to pick and track explicit reference points. If there isn’t enough consistent signal from frame to frame, the alignment will fail and could easily produce the symptoms you’re seeing here. There isn’t really a way to fine-tune this. There isn’t really a way to fine-tune this, but I agree with the supposition by leetleyang and rbs_sci that this micrograph is probably empty.

Hope this helps,