Hi Everyone,
I think I have good 2D classes but when I start an ab initio or NU refinement I can barely see secondary structures like TM helices. My protein is very small, around 60A in diameter, around 150A with micelle.
Could someone give me any advice on how to proceed?
Parameters for 2d classification
Number of 2D classes-200
Initial classification uncertainity factor-2
Circular mask diameter- 180
Number of final full iterations-2
Number of online-EM iterations-40
batchsize per class-1000
For ab initios I started with 8 classes with max resoluition 10 and initial resolution 15 and eventually to 3 classes with max resolution 5 and initial resolution 7.
When I look at the volumes in ab initio I can barely see TM helices.
I can’t see any convincing transmembrane helices in the 2D either, except for possibly (possibly!) in two classes.
I’d try increasing the uncertainty factor for the 2D to 8-10 and running again, perhaps with a target res of 4 rather than 6, perhaps also with a tighter mask around the 2D. Or enable non-negativity and clamp solvent…
@sshrestha for SPA of solely integral membrane proteins you should have some extra cellular domain/fiducial attached to your tmd’s. As you can see micelles here which probably have your protein but with low s/noise ratio. For this strategy to be successful you’ll need very high number of particles for image alignment.
Had i been in your case, i’d rather try to engineer some fiducial protein/ fins a Nb to the tmd’s for image alignment as that makes data processing easier. Rest, you can use clamp solvent and non-negativity in your 2d and see if that improves.
Best.
