I am working with a dataset that was collected super-resolution with pixel size 0.4125. We imported as such and then, during motion correction, F-cropped to 1/2 to arrive at the real pixel size of 0.825. Then, we used the Micrograph Denoiser job, and manually picked 500 particles from a range of defocus values. We extracted with a box size of 256 and F-cropped to 128. No issues so far.
We then try to run topaz train, and get the following error in our log:
WARNING: 395 particle coordinates are out of the micrograph dimensions. Did you scale the micrographs and particle coordinates correctly?
The job eventually completes, but the training fails and produces a flat precision graph. Are we doing something wrong here?
Hi @KDAmico94,
If you are using our the “Micrograph Denoiser” job then the warning you are getting is due to the difference in scaling applied by our Micrograph Denoiser and the scaling applied by TOPAZ. As of v4.5, denoised micrographs produced by our Micrograph Denoiser in CryoSPARC are not compatible with TOPAZ training and extraction. You can technically use our denoised micrographs in TOPAZ (ie the job will launch), but this performs very poorly (“flat curves” on the average-precision).
If you would like to use denoised micrographs for TOPAZ training and extraction you would need to use the TOPAZ Denoise job to generate denoised mics.
Best,
Kye
Hi Kye, thank you so much for the reply. This was immensely helpful.
1 Like