Subset Particles by Statistic

Troubleshooting
1

Hi :distorted_face: , i recently be super confused by the parameter: Minimize over per-particle scale

  1. will this interfere the downstream job? like RBMC or local refinement? should I use this before or later than RBMC. I’ve been received different comment from different AIs.

  2. maybe it has downstream influence? but when I run job - Subset Particles by Statistic and try to split by manual thresholding, it will report an error, so it’s not written in particles’information? then how the downstream has a parameter called refine_scale_reset?

image
image341Ă—231 3.86 KB

image

3

Hi @J6H55!

In this case, we mean minimize in the Bayesian sense. You can imagine this parameter means “Find the per-particle scale which results in the lowest error between the image and the reference”. We recommend this parameter is usually left on for refinements of the whole target, and usually turned off for local refinements of small regions. Once you run it a single time, the particles’ per-particle scales are stored with the particles for downstream jobs. It should definitely be turned on at least once before jobs like RBMC.

The error you’re seeing there is unexpected: when you select “Split by manual thresholding” the “Number of Gaussians” parameter should be replaced by “Number of thresholds”. If you create a new job do you continue to see this behavior?

image
image748Ă—542 24.2 KB

4

image
image351Ă—624 15.4 KB

I already enabled PPS in the last NU-job, but later subset still doesn’t work.
It looks like this

image

5

image
image317Ă—266 8.47 KB

6

image
image538Ă—378 42.9 KB

image
image354Ă—332 10.6 KB

is there any information produced in particle that I can see after I did PPS, I didn’t see changes after I did PPS, as showed above :slightly_frowning_face:

7

What were you trying to do with it? It is intended to split particles into different groups. did it do that?

8

just try to split particles into different groups based on their PPS number, to see if they contribute differently in the entire map
but it reported error, couldn’t start the job

9

PPS number? Not sure what that is?

Oh scale I see

Did you refine scales? Otherwise it won’t be there and will fail

Oh I see during NU - there should be a graph during NU that shows the refined scales - is it there?

10

image
image1206Ă—1068 93.5 KB

I think this is your issue - the subset mode is manual thresholding, but for some reason it shows number of gaussians. What happens if you change the subset mode?

11

yeah, it work fine with split by gaussian

but when I try to use threshold, it reported error, while in rwaldo’s reply it can still work.

my version is cs4.7.1, maybe there’s something wrong with my lab’s cryosparc

12

If it is set to manual thresholding, it shouldn’t display the field for number of gaussians

13

yes

I’ll try the same job in cs5.0beta later, to see if this can be solved