Research Article: Current approaches for automated model building into cryo-EM maps using Buccaneer with CCP-EM

Date Published: June 01, 2020

Publisher: International Union of Crystallography

Author(s): Soon Wen Hoh, Tom Burnley, Kevin Cowtan.


The model-building software Buccaneer has been repurposed based on existing methods for building models into cryo-EM maps. These approaches are implemented in the latest CCP-EM model-building pipeline (version 1.4.0).

Partial Text

Technological advances have brought major improvements in single-particle electron cryo-microscopy (cryo-EM), in particular the introduction of direct electron detectors (McMullan et al., 2016 ▸). This technique has allowed the study of rapidly frozen biological macromolecules without the need for crystallization. The increasing number of cryo-EM maps produced in recent years, especially those of high resolution (<4 Å), has motivated the modification of existing and the development of new model-building tools to interpret cryo-EM maps (Cowtan, 2008 ▸; Baker et al., 2012 ▸; Brown et al., 2015 ▸; Wang et al., 2015 ▸; Chen et al., 2016 ▸; Zhou et al., 2017 ▸; Terwilliger, Adams et al., 2018 ▸; Terashi & Kihara, 2018 ▸; Afonine, Poon et al., 2018 ▸; Chojnowski et al., 2018 ▸; Nicholls et al., 2018 ▸). Unless specified otherwise, all tests were performed on a total of 208 EM maps downloaded from the Electron Microscopy Data Bank (EMDB; Lawson et al., 2011 ▸). The published resolutions of the maps were in the range 1.8–3.97 Å. Map searches were performed on three separate occasions. Criteria such as single particle, protein and a resolution of better than 4.0 Å were used to filter the search results. Maps were excluded if no deposited structure was available at the time of the search and if the volume was too large (for example maps of virus capsids). An initial test was performed to examine the effects of correlation and fast modes on models built by Buccaneer. Results were obtained from one cycle of Buccaneer performed in four different settings without applying any sharpening to the map.(1) Turning on both correlation and fast modes.(2) Turning on only correlation mode.(3) Turning on only fast mode.(4) Turning off both correlation and fast modes. Overall, the adjustments made to Buccaneer and the pipeline settings in the CCP-EM software suite have improved the results from building models into EM maps. Evident improvements in the sequencing step are seen by adapting EM reference structure data in Buccaneer. In general, the performance of Buccaneer in interpreting EM maps decreases as the overall map resolution becomes worse. However, the overall map resolution reported does not necessarily reflect the quality of the map in all regions. An accurately traced fragment can reduce the length required to achieve a success rate of 80% in sequencing by a few residues. More work needs to be performed in order to improve the model-building algorithm in Buccaneer for EM maps. A similar approach will be applied to the Nautilus model-building software for nucleic acids (Cowtan, 2014 ▸). However, more than this might be required for Nautilus to work properly with EM data, as it uses a different algorithm. The overall effectiveness in detecting the ‘fingerprint’ of nucleic acid bases and the sequencing algorithm requires improvement for it to provide better results.   Source:


Leave a Reply

Your email address will not be published.