Re: [scipion-users] Use of masks in MonoRes and LocalDebur

[JOSE L. V. P. <jl...@cn...>]

Dear Matthieu,

Thanks for reporting the issue.

> Is it possible to use a non-spherical mask to restrict where MonoRes and
> LocalDeblur analysis run in a map? I've not seen this option in these
> programs but I'm asking anyhow in case there are some suggestions about how
> to achieve this.

Regarding with MonoRes. The main aspect is that it requires noise to  
determine the resolution. A tight mask to the map allows to  
distinguish the noisy region from the protein, and it defines the  
region for resolution analysis. The spherical mask you comment is  
other issue, it defines the region where noise and protein are. Note,  
that some maps presents noise-free regions outside of a spherical  
region (for instance because particles where masked with a circle or  
the reconstruction used a sphere). The tight mask that MonoRes  
requires only defines the protein.

The spherical mask should not affect to the computational times. I  
think it is due to the your map is large, I guess that probably it  
will be larger than 1000X1000x1000, am I right?. Hence I estimate that  
monores can take 2-4h?. Here is the problem, because internally  
LocalDeBlur runs MonoRes several times and as a consequence it will  
take more time (more than a day, could it be?)

We have a solution for decreasing (a litle bit) MonoRes time and  
therefore LocalDeblur, unfortunately, we are developing it.

> If I could run these programs on a 15% segment of these helical maps (which
> contains all the information), then I could just apply helical symmetry to
> get the full map and the processing would be much faster.
> I suspect that pre-masking the half-maps or giving a mask to MonoRes that
> does not include all the signal of the macromolecule in the map would
> negatively impact the noise estimation in MonoRes.

If I understood right, what you are saying is that due to your map  
presents symmetry, you want to apply MonoRes/LocalDeBlur to the unit  
cell, and then make use of the symmetry to reover the map.
Honestly, I never did it, but you can do to get MonoRes result but is  
not applicable to LocaldeBlur because the current implementation only  
uses a single map. The use of half maps in LocalDeBlur is in our  
development list.
If you want to do it with MonoRes. You  need to create a mask that  
defines the region of your unit cell, and then you provide to monores  
the half maps and that mask. The noise using half maps should only be  
calculated by halves difference.

By the way, what Scipion version are you using? This is important to  
know how fast your MonoRes version is.

I expect to be useful

Thanks again for your questions and comments

Kind regards

Jose Luis Vilas

Matthieu Benoit <bio...@gm...> escribió:

> Hi,
>
> Is it possible to use a non-spherical mask to restrict where MonoRes and
> LocalDeblur analysis run in a map? I've not seen this option in these
> programs but I'm asking anyhow in case there are some suggestions about how
> to achieve this. I'm asking because I have over 20 large maps with helical
> symmetry that I'd like to postprocess. I like the MonoRes/LocalDeblur
> results but the computation takes 2 days for each map on my workstation.

> If I could run these programs on a 15% segment of these helical maps (which
> contains all the information), then I could just apply helical symmetry to
> get the full map and the processing would be much faster.
> I suspect that pre-masking the half-maps or giving a mask to MonoRes that
> does not include all the signal of the macromolecule in the map would
> negatively impact the noise estimation in MonoRes.
>
> Thanks for any advice,
>
> Best,
>
> Matthieu