Re: [scipion-users] 2D masking particles

[Gregory S. <sha...@gm...>]

Hello Teige,

there are a couple of things you should consider: 1) 2d classification in
relion will always start from random angles unless you selected option
"Consider previous alignment" (which in old code would set priors=angles);
2) before applying the mask you should use "xmipp-apply alignment 2d"
protocol to produce a new stack of aligned particles. Alignment assign
protocol will only create a "link" between particle set and aligments.

So the workflow in your case would probably be: relion 3D autorefine ->
relion2D without search (consider prev. alignment=yes) -> (output classes
are therefore oriented as per the 3d map) -> take the side-view class and
create a mask -> create particle subset from relion2D of side-view class ->
xmipp3 apply alignment to subset --> xmipp3 apply 2d mask to the previous
output

Best regards,
Grigory

--------------------------------------------------------------------------------
Grigory Sharov, Ph.D.

MRC Laboratory of Molecular Biology,
Francis Crick Avenue,
Cambridge Biomedical Campus,
Cambridge CB2 0QH, UK.
tel. +44 (0) 1223 267542 <+44%201223%20267542>
e-mail: gs...@mr...

On Thu, Mar 1, 2018 at 4:56 PM, Matthews-Palmer, Teige <
t.m...@im...> wrote:

> Dear Scipion Users,
>
> I have a situation where a domain of my object does not align in 2D or 3D
> (some failed attempts of density subtraction & masked refinement).
> I have side-views where this domain is separate (not overlapping) from the
> rest of the object.
> What I would like to do is to apply a fairly tight 2D mask around this
> domain to the particle images. The xmipp programs in scipion seem to
> provide a convenient way, however I cannot get it to work right yet.
>
> My workflow is: relion 3D autorefine -> relionc2D without search ->
> (output classes are therefore oriented as per the 3d map) -> take the
> side-view class and create a mask -> import mask -> create particle subset
> from relionc2D of side-view class -> xmipp3 apply 2d mask
> The outcome is that the mask is rotated in the output particles, but the
> transformations are wrong (not even close). If I perform relionc2D without
> search on the unmasked side-view particle subset, they do not recreate the
> class average so they must not have the alignments assigned.
>
> If I do instead:
> create particle subset from relionc2D of side-view class -> scipion
> alignment assign using relion 3D autorefine as input alignments -> xmipp
> apply 2d mask
> Then, checking the particle subset, it can recreate the class average
> without searching, so it has the alignments, but the mask is not rotated at
> all in the output particles.
>
> So I tried xmipp3 apply alignment 2d to see if it will actually perform a
> transformation and interpolation on the particle images and create a new
> particle stack, but the output particles are unchanged.
>
> Does anyone have experience of 2D masking and have a workflow that works?
> Any help greatly appreciated.
> All the best,
> Teige
>
>
>
> ------------------------------------------------------------
> ------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
> _______________________________________________
> scipion-users mailing list
> sci...@li...
> https://lists.sourceforge.net/lists/listinfo/scipion-users
>