Wiki Page Content

Q: I was in Houston for your last workshop on EMAn and i would like to know when we can obtain the videos from the course.

They are being converted, but it is a slow process. They should be available on the workshop page when done.

(and now 18 months later, I have no idea what happened :^( )

Q: I am trying to refine a structure with D7 symmetry using Eman1.8. It seemed the sym=d7 did not work at all (even one rough refinement with sym=d7 command resulted in a wrong structure). Is it a bug? Do you have any suggestion? Thank you.

A: Please provide the exact 'refine' command you used (which did not work). startAny doesn't really do symmetry, and startcsym (which is what you should use) does only C7 symmetry. The idea would be to use startcsym with C7 symmetry then use proc3d to impose D7 symmetry on the final result. I think all of this is discussed in the tutorial interface if you tell it you're doing a D7 reconstruction.

Q: Thank you. My refine command that is not working on d7 symmetry: refine 6 ang=15.0 sym=d7 mask=60 pad=128 hard=25 classkeep=1 classiter=8 xfiles=2.54,900,99 phasecls shrink=4 proc=6 median. I am pretty confident that the particles are good (2D averages show clearly d7 symmetry of the particles). I managed to generate starting model with d7 symmetry using eman2 command (e2make3d.py 2d.img --sym=d7 --out=0d.mrc --pad=128). The starting model generated is good. However, after only one iteration of refinement, the structure has already changed a lot (only center part of the particle is remained).

A: Quite a few issues with this refine command.

  • First, ang=15 is MUCH too large for D7 symmetry, and is fairly large even for low symmetry. The largest value one would typically use is ~9 or 10, and even this would rarely be used for D7. 360/14 ~= 25, that means with a 15 degree angular step, you effectively have only 1 (maybe 2) different side-view orientations being sampled at all. I would suggest starting with ~7.5 at the most.
  • I don't have enough details on the project to make exact suggestions in a lot of these parameters, but a few general comments: classiter=8 is fine for 2-3 iterations to get the structure on track, but generally once you have the shape pretty much right, dropping down to 3 is typical. shrink=4 is very aggressive, and as I'm assuming your particle is ~100 pixels in size, this doesn't leave very much detail in the images (ie - the shrunken particles are only ~25 pixels in size) for use in classification.
  • Clearly you aren't doing CTF correction. While this is generally fine, some particles with a high rotational symmetry have a tendency to not converge as well unless you do full CTF correction and use the dfilt= and ctfcw= options. This may or may not be the case here, it is a somewhat rare phenomenon. Generally phasecls will solve the problem sufficiently for low resolution work.
  • Given a mask of 60, I assume your box is ~120. pad=128 isn't much padding for a box size of 120. I would try pad=160.

---Thank you, Steve. I will modify the commands and let you know if they work.

Q:Hello! I am trying to use the pdb2mrc command to convert a DNA initial model (built in Hyperchem and saved as a pdb file) into the required mrc file. Even including het does not result in any of the heteroatoms being incorporated into the final mrc file. Do you have any advice/suggestions/tricks for converting solely DNA based models into mrc files?

A:Sorry, not sure what issue you're having ? What heteroatoms are you trying to include ? pdb2mrc deals only with H,O,C,N,S and P at present. With the 'het' option, it should consider HETATM entries, but if you are including any other element it will be ignored. What else are you trying to put in ?

EMAN1/Projects (last edited 2008-11-26 04:42:30 by localhost)