multi-body refinement is that the dynamics in a macromolecular
complex can be described by a discrete number of independently
moving, rigid bodies. This model builds on the observation that the
structure of individual protein domains, or tertiary structure, often
remains intact upon continuous changes in the overall (quaternary)
structure of flexible complexes. During multi-body refinement, the
signal from all-but-one body is iteratively removed from the experimental images using partial signal subtraction [3–5], and the signal
from the remaining body is aligned with respect to reference projections of only that body. Thereby, alignments of the individual
bodies are not compromised by differences in the relative orientations between the bodies among the images in the data set, and
reconstructions from the individual bodies are better defined than
in a single refinement of the entire data set. Moreover, upon convergence of the multi-body refinement, principal component analysis on the relative orientations of all bodies for every experimental
image in the data set is used to characterize the most dominant
motions in the complex.
This chapter describes how to perform multi-body refinement
of a flexible macromolecular complex in RELION, and how to
interpret the results. We use a previously published data set on the
pre-catalytic spliceosomal B-complex [6] as an example. These data,
including the necessary files to perform multi-body refinement, are
publicly available through the EMPIAR data base [7], under accession number 10180. We describe how to design real-space masks
that define the individual bodies (Fig. 1); we provide detailed
instructions on how to choose the various parameters; and we
describe how to analyze the results by post-processing the individual body reconstructions (Figs. 2 and 3) and by characterizing the
motions through principal components analysis (Fig. 4). A minimum molecular weight for the individual bodies, which is required
for accurate alignment, limits the applicability of this method to
relatively large complexes that can be divided into two or more
rigid bodies of at least 100–150 kDa. Besides this caveat, multibody refinement is generally applicable to flexible macromolecular
complexes.
2 Materials
2.1 Computer
Hardware
Requirements
1. A computer with a reasonably modern CPU with at least
8 cores. We used a computer with a 12-core Intel Xeon
ES-2620v3 CPU at 2.4 GHz. Hyper-threading resulted in
24 visible cores in the operating system.
2. At least 400 GB of disk space to execute the example case
described in this chapter. We used a BeeGFS file system of
146
Takanori Nakane and Sjors H. W. Scheres
complex can be described by a discrete number of independently
moving, rigid bodies. This model builds on the observation that the
structure of individual protein domains, or tertiary structure, often
remains intact upon continuous changes in the overall (quaternary)
structure of flexible complexes. During multi-body refinement, the
signal from all-but-one body is iteratively removed from the experimental images using partial signal subtraction [3–5], and the signal
from the remaining body is aligned with respect to reference projections of only that body. Thereby, alignments of the individual
bodies are not compromised by differences in the relative orientations between the bodies among the images in the data set, and
reconstructions from the individual bodies are better defined than
in a single refinement of the entire data set. Moreover, upon convergence of the multi-body refinement, principal component analysis on the relative orientations of all bodies for every experimental
image in the data set is used to characterize the most dominant
motions in the complex.
This chapter describes how to perform multi-body refinement
of a flexible macromolecular complex in RELION, and how to
interpret the results. We use a previously published data set on the
pre-catalytic spliceosomal B-complex [6] as an example. These data,
including the necessary files to perform multi-body refinement, are
publicly available through the EMPIAR data base [7], under accession number 10180. We describe how to design real-space masks
that define the individual bodies (Fig. 1); we provide detailed
instructions on how to choose the various parameters; and we
describe how to analyze the results by post-processing the individual body reconstructions (Figs. 2 and 3) and by characterizing the
motions through principal components analysis (Fig. 4). A minimum molecular weight for the individual bodies, which is required
for accurate alignment, limits the applicability of this method to
relatively large complexes that can be divided into two or more
rigid bodies of at least 100–150 kDa. Besides this caveat, multibody refinement is generally applicable to flexible macromolecular
complexes.
2 Materials
2.1 Computer
Hardware
Requirements
1. A computer with a reasonably modern CPU with at least
8 cores. We used a computer with a 12-core Intel Xeon
ES-2620v3 CPU at 2.4 GHz. Hyper-threading resulted in
24 visible cores in the operating system.
2. At least 400 GB of disk space to execute the example case
described in this chapter. We used a BeeGFS file system of
146
Takanori Nakane and Sjors H. W. Scheres