Support #4214
open
Added by Bridget Carragher over 8 years ago.
Updated almost 7 years ago.
Category:
Image Processing
Affected Version:
Appion/Leginon 3.2
Description
When making a stack if you select the main menu option of CTF conf > 0.8 doe this currently do anything? I tried instead to use the advanced option but enede up with a zero particle stack.
Also on the summary page if you put in a value for minimum allowed confidence how is this related to the CTF conf. values that are now calculated. This used to be a way to check how many images you would be using if you implemented a CTF conf. cutoff but not sure it works at all anymore?
- Assignee changed from Sargis Dallakyan to Bridget Carragher
Issue One, please provide session name and runname to point to the specific case. Reading the code it should still work.
Issue Two, confidence and confidence_d is no longer used. Neil's fit-resolution is in general used but he will have to answer what should do into the discremination algorithm.
OK, well I was trying to do this in:
http://192.168.4.37/betamyamiweb/processing/checkAppionJob.php?expId=1909&jobId=623
I am not sure if the session and run are the issue. It is more the question of whether the "ctf confidence" value is still well defined in the several places that it resides. It has been redefined I think by the new CTF estimation and I am not sure this has migrated over to the various other places: e.g. the summary page where we can restrict this value and see how many images make it through; the stack making launch where you can set the stack criterion in several ways.
I was trying to review this and I may be lost.
Is the problem that SPIDER CTF correction does not work on rectangle images. My guess is that ACE2 would fail as well. What do you guys use to CTF correct rectangle images? Do we need to add more functions to makestack?
I personally do not use the CTF confidence any more and just look at the resolution cutoff, but it is a good default setting. I guess I would have to look at your data and setup more to see what is going on.
I do not know what works with rectangular images but I think we need to let users know what does and does not work - maybe just in the pop up help for now.
We chatted at group meeting and all agreed that the default 0.8 confidence value does not make sense anymore. I agree that we should use the range cutoffs instead. But I think the language on the popups could be refined to make this more clear and I think this should be brought up to the main menu (i..e out of advanced settings).
I would also request that the same query could be run on the CTF summary pages so that you can get an idea of how many images are passing the criteria. Or better yet have this query in the ACE menu on the viewer.
It would be a delightful feature if we could suggest values for the CTF cutoff limits based on the instrument that is being used. E.g. if it is a Krios then your probably want to have a cutoff of 8A or better, a mid-range FEG, 10A or better and a Lab6 etc. a cutoff of 20A or nothing.
If Neil or someone else goes into the code to fix the CTF confidence issue (i.e. get ris of the old one and move the new method out of advanced) I think that the popup docs for the confidence inputs should say something like:
"Resolution at 0.8 confidence better than: [insert number] A
same for 0.5
I had a MySQL query that would look at the CTF resolution values for the data set and then suggest a cutoff, but it was causing you guys to have really really slow page load times, so I commented it out.
Sounds like a useful thing but could it be run in the background as a stand alone job that does not slow down page loading?
Maybe worth adding the ctf histogram interface that is in ctfreport.php in this page so that people can try out different criteria.
Also available in: Atom
PDF