Chapter 9
Automated Modeling and Validation of Protein Complexes
in Cryo-EM Maps
Tristan Cragnolini, Aaron Sweeney, and Maya Topf
Abstract
The resolving power of cryo-EM experiments has dramatically improved in recent years. However, many
cryo-EM maps may still not achieve a resolution that is sufficiently high to allow model building directly
from the map. Instead, it is common practice to fit an initial atomic model to the map and refine this model.
Depending on the resolution and whether the structure suffers from inherent flexibility or experimental
limitations, different methods can be applied, to obtain high-quality, well-fitted atomic model of the
macromolecular assembly represented by the map, and to assess its properties. In this review, we describe
some of these methods, with the main focus on those that have been developed in our group over the last
decade.
Key words Cryo-EM, Assembly, Density fitting, Validation, Structure modeling
1 Introduction
The resolving power of cryo-EM experiments has dramatically
improved in recent years, now rivalling crystallography, with nearatomic resolutions for many systems. Technical advances, in particular, direct electron detectors and better image processing methods
have made it possible to obtain resolutions better than 3.5 A ˚ in
some cases, making cryo-EM a popular tool to probe the structure
of macromolecular assemblies, even for low purity samples. Other
related methods, such as cryo-ET and subtomogram averaging can
elucidate the structure of complex samples, entire cell components,
tissues, or full viral particles.
However, the relatively low resolution in still most cryo-EM
maps limits their applicability to model building based on the
information from the map only. Instead, it is common practice to
fit an initial atomic model into the density map and refine it. These
initial models can either be experimentally derived (e.g., from X-ray
crystallography, NMR spectroscopy, or high-resolution cryo-EM),
or comparative models (homology models build from sequence
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_9, © Springer Science+Business Media, LLC, part of Springer Nature 2021
189
Automated Modeling and Validation of Protein Complexes
in Cryo-EM Maps
Tristan Cragnolini, Aaron Sweeney, and Maya Topf
Abstract
The resolving power of cryo-EM experiments has dramatically improved in recent years. However, many
cryo-EM maps may still not achieve a resolution that is sufficiently high to allow model building directly
from the map. Instead, it is common practice to fit an initial atomic model to the map and refine this model.
Depending on the resolution and whether the structure suffers from inherent flexibility or experimental
limitations, different methods can be applied, to obtain high-quality, well-fitted atomic model of the
macromolecular assembly represented by the map, and to assess its properties. In this review, we describe
some of these methods, with the main focus on those that have been developed in our group over the last
decade.
Key words Cryo-EM, Assembly, Density fitting, Validation, Structure modeling
1 Introduction
The resolving power of cryo-EM experiments has dramatically
improved in recent years, now rivalling crystallography, with nearatomic resolutions for many systems. Technical advances, in particular, direct electron detectors and better image processing methods
have made it possible to obtain resolutions better than 3.5 A ˚ in
some cases, making cryo-EM a popular tool to probe the structure
of macromolecular assemblies, even for low purity samples. Other
related methods, such as cryo-ET and subtomogram averaging can
elucidate the structure of complex samples, entire cell components,
tissues, or full viral particles.
However, the relatively low resolution in still most cryo-EM
maps limits their applicability to model building based on the
information from the map only. Instead, it is common practice to
fit an initial atomic model into the density map and refine it. These
initial models can either be experimentally derived (e.g., from X-ray
crystallography, NMR spectroscopy, or high-resolution cryo-EM),
or comparative models (homology models build from sequence
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_9, © Springer Science+Business Media, LLC, part of Springer Nature 2021
189