 |
 |
Hi Craig, In that case I will opt for a less aggressive subtraction of the micelle, and see what happens. Best wishes, Ahmed On 4 May 2017, at 17:33, Craig Yoshioka <[log in to unmask]> wrote: > Hi Ahmed, > > I have tried this many times, with mixed results. I think the biggest issue is the particle normalization becomes very critical, especially when you are trying to subtract a micelle away from TMDs. Also, when subtracting the micelle from a TMD, there is a greater potential for introducing bias than with normal subtraction. > > The first step I usually try is to subtract the micelle region (including TMD) capped at ~ average micelle density. I did a lot of this in Chimera with a combination of deleting everything but the micelle, low-pass filtering, and thresholding. This also subtracts some density from the TMD, but at least lessens the influence of the micelle on alignment and classification, and does not overly bias the remaining TMD density. After the first subtraction I sometimes would try more aggressive subtractions, but that’s where you start seeing more problems. I also check the reconstruction with the original particles using the angles from the subtracted particles at every step, as a sanity check. It’s a good sign, for example, to see the micelle get a little blurrier and the TMD get a little sharper. > > >> On May 3, 2017, at 2:52 PM, Ahmed-Noor Adam Agip <[log in to unmask]> wrote: >> >> Hi, >> >> >> I understand particle subtracted experimental images in relion can aid in 3D classification, but is it valid to use the very same subtracted experimental images to refine the structure? For example, if I plan to subtract away part of the detergent belt of my protein, in order to align my protein better during 3D refinement? >> >> >> Best wishes, >> Ahmed >