search

rsanchezgarc / deepEMhancer

Deep learning for cryo-EM maps post-processing
Apache License 2.0
49 stars 8 forks source link

Are there some referenced parameters that can indicate the quality of map improvement based on different regions? #24

Open JuneHanzhou opened 1 year ago

JuneHanzhou commented 1 year ago

Hi, dear authors, I have been used the DeepEMhancer for a long time. It works well in overall map improving. But there are some questions I am confused about: When I deal with new structures, since it can improve the EM-map, I feel curious and uncertain about which parts of the improved map are reliable? To what extent can this result be trusted? Is there any referenced indicators or data? Since the oversharpened regions can sometimes be observed. Are there some referenced parameters that can indicate the quality of map improvement based on different regions. Thanks sincerely.

rsanchezgarc commented 1 year ago

Hi,

This is quite an important question. Unfortunately, since new proteins have no ground truth to compare against, the only way we can say something about the results is consistency, and of course, consistency does not imply correctness. Having said that, what we recommend is to compare the results obtained with several post-processed maps. I know this is time-consuming, but it is the safest way. I cannot think of a simple method that gives you a number.

I also like the approach they follow in this paper https://doi.org/10.1128/jvi.00282-21, in which they focus on the regions they are interested in, and then, they try to find a local b-factor that works as well as deepEMhancer.