Abstract
Cryo-EM maps are valuable sources of information for protein structure modeling. However, due to the loss of contrast at high frequencies, they generally need to be post-processed to improve their interpretability. Most popular approaches, based on global B-factor correction, suffer from limitations. For instance, they ignore the heterogeneity in the map local quality that reconstructions tend to exhibit. Aiming to overcome these problems, we present DeepEMhancer, a deep learning approach designed to perform automatic post-processing of cryo-EM maps. Trained on a dataset of pairs of experimental maps and maps sharpened using their respective atomic models, DeepEMhancer has learned how to post-process experimental maps performing masking-like and sharpening-like operations in a single step. DeepEMhancer was evaluated on a testing set of 20 different experimental maps, showing its ability to reduce noise levels and obtain more detailed versions of the experimental maps. Additionally, we illustrated the benefits of DeepEMhancer on the structure of the SARS-CoV-2 RNA polymerase.
Highlights
Cryo-EM maps are valuable sources of information for protein structure modeling
In order to assess the quality of DeepEMhancer predictions, we first compared them against the target maps generated by LocScale
For DeepEMhancer maps, we measured a median correlation coefficient of 0.9 against LocScale maps in contrast to 0.6 for input maps. Such an important increase in the correlation coefficient implies that DeepEMhancer has learned to accurately reproduce the effect of LocScale sharpening with one important advantage: no atomic models are required to employ DeepEMhancer
Summary
Cryo-EM maps are valuable sources of information for protein structure modeling. due to the loss of contrast at high frequencies, they generally need to be post-processed to improve their interpretability. Most popular approaches, based on global B-factor correction, suffer from limitations They ignore the heterogeneity in the map local quality that reconstructions tend to exhibit. The first sharpening approach for cryo-EM maps was introduced by Rosenthal and Henderson[1] and their formulation, based on the global B-factor correction, is still at the basis of the most commonly employed sharpening methods, including RELION postprocessing[2,3] or Phenix AutoSharpen[4] The principle behind these algorithms consists in the correction of the raw maps by boosting the amplitude of their high-frequency Fourier components. LocSpiral[8] employs the spiral phase transformation to factorize the volume and perform a local enhancement based on the normalization and thresholding of the amplitudes Despite their benefits, current local sharpening approaches present some drawbacks. The main strength of LocScale, its ability to employ the structural information of atomic models, could be regarded as its main weakness since the availability of atomic models limits its applicability
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
