Chapter 8
Evaluating Local and Directional Resolution of Cryo-EM
Density Maps
Sriram Aiyer, Cheng Zhang, Philp R. Baldwin, and Dmitry Lyumkis
Abstract
A systematic and quantitative evaluation of cryo-EM maps is necessary to judge their quality and to capture
all possible sources of error. A single value for global resolution is insufficient to accurately describe the
quality of a reconstructed density. We describe the estimation and evaluation of two additional resolution
measures, local and directional resolution, using methods based on the Fourier shell correlation (FSC). We
apply the protocol to samples that encompass different types of pathologies a user is expected to encounter
and provide analyses on how to interpret the output files and resulting maps. Implementation of these tools
will facilitate density interpretation and can guide the user in adapting their experiments to improve the
quality of cryo-EM maps, and by extension atomic models.
Key words Anisotropy, Preferred orientation, Fourier shell correlation, Sampling, Cryo-EM, Singleparticle analysis
1 Introduction
In a cryo-EM experiment, images of frozen hydrated macromolecules are recorded using an electron microscope. Individual macromolecular “particles” embedded within the layer of vitreous ice can
be distributed in different orientations relative to the electron
beam. A computational assignment of their orientation distributions is necessary to perform a 3D reconstruction of the imaged
object. This gives rise to a reconstructed 3D density map. A number of recent reviews describe the various procedures in detail [1–
3]. The quality of the reconstructed density is then assessed using a
global resolution metric.
A single global resolution metric may not be sufficient to
ensure that a map can be properly interpreted with an atomic
model. Several other resolution measures should be satisfied to
better interpret a map. We discuss here two such measures. The
first is a windowed, or “local resolution” evaluation approach [4],
where sections of final half-maps are compared using the usual
Tamir Gonen and Brent L. Nannenga (eds.), CryoEM: Methods and Protocols, Methods in Molecular Biology, vol. 2215,
https://doi.org/10.1007/978-1-0716-0966-8_8, © Springer Science+Business Media, LLC, part of Springer Nature 2021
161
Evaluating Local and Directional Resolution of Cryo-EM
Density Maps
Sriram Aiyer, Cheng Zhang, Philp R. Baldwin, and Dmitry Lyumkis
Abstract
A systematic and quantitative evaluation of cryo-EM maps is necessary to judge their quality and to capture
all possible sources of error. A single value for global resolution is insufficient to accurately describe the
quality of a reconstructed density. We describe the estimation and evaluation of two additional resolution
measures, local and directional resolution, using methods based on the Fourier shell correlation (FSC). We
apply the protocol to samples that encompass different types of pathologies a user is expected to encounter
and provide analyses on how to interpret the output files and resulting maps. Implementation of these tools
will facilitate density interpretation and can guide the user in adapting their experiments to improve the
quality of cryo-EM maps, and by extension atomic models.
Key words Anisotropy, Preferred orientation, Fourier shell correlation, Sampling, Cryo-EM, Singleparticle analysis
1 Introduction
In a cryo-EM experiment, images of frozen hydrated macromolecules are recorded using an electron microscope. Individual macromolecular “particles” embedded within the layer of vitreous ice can
be distributed in different orientations relative to the electron
beam. A computational assignment of their orientation distributions is necessary to perform a 3D reconstruction of the imaged
object. This gives rise to a reconstructed 3D density map. A number of recent reviews describe the various procedures in detail [1–
3]. The quality of the reconstructed density is then assessed using a
global resolution metric.
A single global resolution metric may not be sufficient to
ensure that a map can be properly interpreted with an atomic
model. Several other resolution measures should be satisfied to
better interpret a map. We discuss here two such measures. The
first is a windowed, or “local resolution” evaluation approach [4],
where sections of final half-maps are compared using the usual
Tamir Gonen and Brent L. Nannenga (eds.), CryoEM: Methods and Protocols, Methods in Molecular Biology, vol. 2215,
https://doi.org/10.1007/978-1-0716-0966-8_8, © Springer Science+Business Media, LLC, part of Springer Nature 2021
161