|
Size: 504
Comment:
|
← Revision 8 as of 2010-08-26 17:38:27 ⇥
Size: 1060
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 3: | Line 3: |
| '''A:''' Yes, but the stacks must already have been aligned and have their Euler angles stored in the image headers. If you haven't done this you can use IMOD to perform alignment and then convert the output into a format suitable for eman2 using e2ali2img.py. Then you can reconstruct the data using the zsample parameter in the fourier reconstructor. There will be more information available on this aspect of eman2 as the source code matures. | '''A:''' Yes, it is now possible to directly reconstruct aligned stacks (.ali) files from imod using the corresponding .tlt file. In fact, the Fourier reconstruction algorithm used in EMAN2 has produced substantially better results than either back-projection or SIRT in a number of cases. Run e2make3d.py --help for more options. Pay particular attention to the various parameters impacting reconstruction size, because the default would be to make a cube using the maximum dimension. ie - if you reconstructed a 2k x 2k stack using default options, it would try to do a 2k x 2k x 2k volume !!! (a very bad thing). You may also want to consider using a mode other than the default 2-pixel interpolation in the Fourier reconstruction. For sparsely sampled tomography data, using a larger interpolation kernel can produce structures which are better localized on the Z axis. However, this will also cause some radial intensity dampening in the final reconstruction (which can be corrected). |
Q: Does e2make3d.py work on tomographic stacks?
A: Yes, it is now possible to directly reconstruct aligned stacks (.ali) files from imod using the corresponding .tlt file. In fact, the Fourier reconstruction algorithm used in EMAN2 has produced substantially better results than either back-projection or SIRT in a number of cases.
Run e2make3d.py --help for more options. Pay particular attention to the various parameters impacting reconstruction size, because the default would be to make a cube using the maximum dimension. ie - if you reconstructed a 2k x 2k stack using default options, it would try to do a 2k x 2k x 2k volume !!! (a very bad thing).
You may also want to consider using a mode other than the default 2-pixel interpolation in the Fourier reconstruction. For sparsely sampled tomography data, using a larger interpolation kernel can produce structures which are better localized on the Z axis. However, this will also cause some radial intensity dampening in the final reconstruction (which can be corrected).