orientation. Such artifacts make de novo modeling more challenging
and could lead to incorrect map interpretation. Although tilting
ameliorates the effects of artifactual features, the problem of preferred orientation is not necessarily completely overcome because
there is still a missing cone that accounts for residual elongation and
loss of resolution in the corresponding direction within the resulting map (see Fig. 9b). Importantly, it is not necessary to have a
perfectly isotropic map in order to accurately interpret the structural features. In fact, even with moderate levels of anisotropy, the
map may appear normal to the eye, and can be used for model
building and refinement. However, care must be taken during the
interpretation of the map and the atomic model, in order to make
sure that structural features are arising from the true features in the
imaged macromolecular object, and not due to artifacts caused by
loss of directional resolution.
A combination of the analyses described above can guide the
user in quantitatively characterizing the directional resolution of an
EM map and thus altering strategies for improving resolution and
overall map quality.
4 Notes
1. FSC-based methods, similar to the one described in the current
protocol, are also implemented in most of the major processing
packages, including Relion [28] and cryoSparc [29]. A popular
alternative approach to the FSC-based analysis is the ResMap
approach, which assigns significance to features within an
experimental density map if and only if a 3D local sinusoid of
a nominal wavelength is statistically detectable above the noise
level [30]. The ResMap approach only requires a single map as
input. Another recent alternative, MonoRes [31], is implemented within the Xmipp/Scipion processing packages
[32, 33]. MonoRes is based on the identification of a monogenic amplitude across different spatial frequencies in a map
and determining whether, at the specific resolution and location in the map, this amplitude is significantly higher than what
is expected from pure noise. MonoRes also only requires a
ä
Fig. 9 (continued) nitrogen in blue, oxygen in red and sulfur in green). All three views show uniform map
density allowing for unambiguous interpretation during model building. (b). HA trimer, first row map from
0
untilted images (map color grey) and second row the same map overlaid with an atomic model (blue). Map
features appear stretched, especially when rotated across the X and Y axes, resulting in artifactual density.
The third (map only) and fourth rows (model overlayed into map at 50% transparency) show a reconstruction
from 40
tilted images. Now, the map features appear visibly better with less stretching of map density. HA
reconstructions depict a β sheet region. Map and model for both AAV2 and HA were aligned using the fit
function in UCSF Chimera [14]
Local and Directional Resolution in Cryo-EM Maps
179
and could lead to incorrect map interpretation. Although tilting
ameliorates the effects of artifactual features, the problem of preferred orientation is not necessarily completely overcome because
there is still a missing cone that accounts for residual elongation and
loss of resolution in the corresponding direction within the resulting map (see Fig. 9b). Importantly, it is not necessary to have a
perfectly isotropic map in order to accurately interpret the structural features. In fact, even with moderate levels of anisotropy, the
map may appear normal to the eye, and can be used for model
building and refinement. However, care must be taken during the
interpretation of the map and the atomic model, in order to make
sure that structural features are arising from the true features in the
imaged macromolecular object, and not due to artifacts caused by
loss of directional resolution.
A combination of the analyses described above can guide the
user in quantitatively characterizing the directional resolution of an
EM map and thus altering strategies for improving resolution and
overall map quality.
4 Notes
1. FSC-based methods, similar to the one described in the current
protocol, are also implemented in most of the major processing
packages, including Relion [28] and cryoSparc [29]. A popular
alternative approach to the FSC-based analysis is the ResMap
approach, which assigns significance to features within an
experimental density map if and only if a 3D local sinusoid of
a nominal wavelength is statistically detectable above the noise
level [30]. The ResMap approach only requires a single map as
input. Another recent alternative, MonoRes [31], is implemented within the Xmipp/Scipion processing packages
[32, 33]. MonoRes is based on the identification of a monogenic amplitude across different spatial frequencies in a map
and determining whether, at the specific resolution and location in the map, this amplitude is significantly higher than what
is expected from pure noise. MonoRes also only requires a
ä
Fig. 9 (continued) nitrogen in blue, oxygen in red and sulfur in green). All three views show uniform map
density allowing for unambiguous interpretation during model building. (b). HA trimer, first row map from
0
untilted images (map color grey) and second row the same map overlaid with an atomic model (blue). Map
features appear stretched, especially when rotated across the X and Y axes, resulting in artifactual density.
The third (map only) and fourth rows (model overlayed into map at 50% transparency) show a reconstruction
from 40
tilted images. Now, the map features appear visibly better with less stretching of map density. HA
reconstructions depict a β sheet region. Map and model for both AAV2 and HA were aligned using the fit
function in UCSF Chimera [14]
Local and Directional Resolution in Cryo-EM Maps
179