Dear Huy and Craig,
The below will fix the priors to always do local searches around those. If
you do not do this, then the local searches will be around the optimal
angles (and translations) from the previous iteration, and will thus be
able to slowly move away.
About starting 3D auto-refinements from local searches, by setting
"perform local searches from" equal to the initial sampling rate: this did
not function as expected in relion-1.3 (which would start by exhaustive
searches anyway). This has been fixed in relion-1.4, so if you set it this
way, local searches will be done from the first iteration.
We very often pass the data.star from one 3D classification/auto-refine
job into the next.
HTH,
Sjors
> Dear Craig,
>
> For Refinement, what I did is to modify in the star file
>
> _rlnOriginX
> _rlnOriginY
> _rlnOriginZ
> _rlnAngleRot
> _rlnAngleTilt
> _rlnAnglePsi
>
> to
>
> _rlnOriginXPrior
> _rlnOriginYPrior
> _rlnOriginZPrior
> _rlnAngleRotPrior
> _rlnAngleTiltPrior
> _rlnAnglePsiPrior
>
> And in the refinement, I use the opition --sigma_angle 1.5 to limit the
> search angle (+/- 3 x 1.5)
>
> I saw this from an older discussion, though not so elaborated by Sjors but
> it seems to work well for me.
>
> Best,
> Huy
>
--
Sjors Scheres
MRC Laboratory of Molecular Biology
Francis Crick Avenue, Cambridge Biomedical Campus
Cambridge CB2 0QH, U.K.
tel: +44 (0)1223 267061
http://www2.mrc-lmb.cam.ac.uk/groups/scheres
|