Local resolution estimation not working on homogenous refined map (refined on an external mask)

Hello Everyone,
I am struggling to estimate the local resolution of a map which is homogenous refined on an external mask (Mask created on a homologous model of protein in Chimera and then imported in CS). This final refined map, when used to estimate Local resolution, its not generating any output, although the jobs gets completed. I tried changing the kernel width in voxels as well as step size but the same problem persists. I am attaching the snapshots of the interface.Screenshot_2020-10-22 J1842 (P7 W11) cryoSPARC v2 Screenshot_2020-10-22 J1842 (P7 W11) cryoSPARC v2(1) .

I just wanted to add one more problem in continuation of the above query. So when the homogenous refined map with an external mask (Mask created on a homologous model of protein in Chimera and then imported in CS) is subjected to Non-Uniform refinement also, the NU-Ref also throws the same error every time(although the job runs fine till 5-6 iterations). So i thought these 2 problems are linked when using Inputs which refined from external masks.Screenshot_2020-10-22 J1858 (P7 W11) cryoSPARC v2

Hi @SangeetaNir,

The first image indicates that the mask range is between 0 and 0.056, but cryoSPARC expects masks to be valued between 0 and 1. For the Local Resolution and Non-Uniform refinement jobs, it is necessary to first pre-process the mask by passing it through it to a Volume Tools job. This will threshold the mask (converting it to a [0,1] binary mask), and optionally, dilate/pad it. A set of appropriate parameters for the volume tools job would look like the following (note that the “Type of input/output volume” parameter must be set to mask):

Screen Shot 2020-10-22 at 5.07.15 PM

The outputted mask from the volume tools job should be used for any further processing (local resolution estimation, refinement, etc.). It’s also recommended to apply a nonzero “Soft padding width” to the mask, especially if it is used in refinement. This will reduce Fourier-space artefacts and make sure that the mask doesn’t harshly clip any of the density. A good value for this is at least 4-5 (pixels).

Best,
Michael

2 Likes

Thank you Michael for your help. I really appreciate that you simplified the problem to make me understand it a lot better. It worked. :slight_smile:

3 Likes