conformations as demonstrated for the MloK1 potassium channel
dataset. In summary, this method presents a viable alternative to
conventional 2D crystallography, when the 2D crystal form is
related to protein function, or when 2D crystals were intentionally
designed.
4 Notes
1. For other possible hardware examples, see the official online
FOCUS documentation: http://focus.c-cina.unibas.ch/wiki/
doku.php?id¼1_0:hardware_linux or http://focus-em.org
2. Depending on which packages you use, GPU cards may not be
required. Please consult the specific documentation of the
software packages present in your data processing pipeline.
3. We have also successfully used cisTEM [32], RELION [52]
and cryoSPARC [31] with 2D crystal data, although these
packages have not been as extensively tested and customized
as frealign-2dx in this context.
4. Actually, just merging in 2D may be enough in some cases, as
long as phase origins are sufficiently accurate. In this case,
however, an initial 3D map needs to be obtained by other
means. For example, we have successfully used stochastic gradient descent (SGD) [31] to this end.
5. It is assumed the dataset has been pruned accordingly before
and/or during the conventional 2D crystal reconstruction
workflow.
6. The program par2star.py can also be called from the command
line as a standalone tool.
7. If using cryoSPARC, please note that, at present, it ignores
these fields and always randomizes the order of the particles
when assigning its half-sets. In this case, it is up to the user to
ensure that the resolution estimation is unbiased by, for example, re-splitting the data based on the 2D crystals after the
refinement and calculating half-maps using FREALIGN or
RELION.
8. If using RELION, similar behavior would be accomplished by
using local searches, i.e., setting the initial angular sampling
equal to the sampling from which to perform local searches.
9. For the signal subtraction and 3D classification tasks, the user
should work from the command line, as these options are not
available from the FOCUS GUI.
Single Particle Analysis for High-Resolution 2D Electron Crystallography
281
dataset. In summary, this method presents a viable alternative to
conventional 2D crystallography, when the 2D crystal form is
related to protein function, or when 2D crystals were intentionally
designed.
4 Notes
1. For other possible hardware examples, see the official online
FOCUS documentation: http://focus.c-cina.unibas.ch/wiki/
doku.php?id¼1_0:hardware_linux or http://focus-em.org
2. Depending on which packages you use, GPU cards may not be
required. Please consult the specific documentation of the
software packages present in your data processing pipeline.
3. We have also successfully used cisTEM [32], RELION [52]
and cryoSPARC [31] with 2D crystal data, although these
packages have not been as extensively tested and customized
as frealign-2dx in this context.
4. Actually, just merging in 2D may be enough in some cases, as
long as phase origins are sufficiently accurate. In this case,
however, an initial 3D map needs to be obtained by other
means. For example, we have successfully used stochastic gradient descent (SGD) [31] to this end.
5. It is assumed the dataset has been pruned accordingly before
and/or during the conventional 2D crystal reconstruction
workflow.
6. The program par2star.py can also be called from the command
line as a standalone tool.
7. If using cryoSPARC, please note that, at present, it ignores
these fields and always randomizes the order of the particles
when assigning its half-sets. In this case, it is up to the user to
ensure that the resolution estimation is unbiased by, for example, re-splitting the data based on the 2D crystals after the
refinement and calculating half-maps using FREALIGN or
RELION.
8. If using RELION, similar behavior would be accomplished by
using local searches, i.e., setting the initial angular sampling
equal to the sampling from which to perform local searches.
9. For the signal subtraction and 3D classification tasks, the user
should work from the command line, as these options are not
available from the FOCUS GUI.
Single Particle Analysis for High-Resolution 2D Electron Crystallography
281