SPRING Stalling Error on Segment and Michelixtrace Jobs

[Colin Deniston]

Hello,

I've been trying to run a couple different jobs on some data today and have been running into a problem when the run seems to get stalled indefinitely without throwing any kind of error. I tried to simplify the issue by running Michelixtrance in a folder containing only a single .mrc image file and a single .hdf reference file but still ran into the same image. I've attached two images to help show what I'm seeing. One image is of my SPRING GUI with all inputs when I start the run and the second image is of the last line of output I get before my run stalls.

I've also seen this same issue when I run on different computers, on different datasets, and during Segment or Michelixtrace jobs. They all have the same problem. However, I can successful run a Micctfdetermine job without any issues. Any advice on where this issue could be arising would be of great help.

Thanks again,
Colin Deniston

[Carsten Sachse]

Dear Colin,

Thank you for posting your issue and providing screenshots. I could see that you are still using Spring v0.85. In order to take full advantage of the MICHELIXTRACE features as they were published in J Struc Biol recently, please update to version 0.86. This version includes the improved statistical thresholding and also pruning based on an expected population of persistence lengths. 

If you are still finding stalling issues with stalling using SEGMENT, could you provide me with more details (parameters and also examples of hdf and box files) to debug the issue? Thank you.

Best wishes,

Carsten 

[Colin Deniston]

Thanks for the suggestions. I'm getting it updated now and will let you know. If I still have problems I'll send another replay with more details on the problem.

[Colin Deniston]

Not sure if this email sent properly so I'm trying again.

To Carsten,

Hello again. So I finally got the most recent version of SPRING installed and have verified that the issue is still occurring. To make it easy for you we can start trouble shooting with the small TMV dataset you provide on your website. I downloaded the first image and box file (TMV_Krios_FalconII_001.mrc and TMV_Krios_FalconII_001.box respectively). When I have them in their own folder and try to run a simple segment run using all the inputs given in your tutorial (segment size 350A, helix width 190A, step size 70A, and pixel size 1.062A) under Spring 2d I run into the same issue I see with my own data when I run either segment or michelixtrace. I have attached the .log file and parameter file from the most recent attempt. I have also included a list of modules I have loaded on my machine when I run SPRING in case that might be helpful. Let me know if you need any other information about my setup. Thanks again!

Colin Deniston
UCSD

On Friday, January 5, 2018 at 11:14:10 PM UTC-8, Carsten Sachse wrote:

[Carsten Sachse]

Hi Collin,

Thanks for being persistent. We reproduced your issue. It is related to the beginner parameter choices that rely on defaults. In your case, CTF correction was turned on silently by default. This option, however, requires a spring.db path. For mpi runs on workstations sometime the errors are not correctly displayed.

We attach a corrected parameters file that will work. 

You can invoke it by:

% segment --f parameters_suggestion.par --d parameters_suggestion_run

In the next release, we will make this more transparent.

Best wishes,

Carsten

[Colin Deniston]

Thanks for the help! Sadly it seems it still runs into the same issue for me! I've attached the report file when I run your parameter file using the command you gave. I've also attached an image of the other output from the stalled run; the contents of the TMV_Krios_FalconII_001 directory inside the parameters_suggestion_run output directory, in case it helps.

I also noticed one other thing that might help. Since you mentioned the CTF being an issue I tried to run Micctfdetermine on the same TMV micrograph with the suggested spherical aberation (2.7), voltage (300), and pixel size (1.062) listed on the website and surprisingly that had the exact same stalling error. However, when I run my own micrograph with my own microscope parameters it runs fine! I can try and send that micrograph to you if you'd like to see it but it's likely too big to be attached via email. So in lieu of it I've sent the report for the failed ctf run on the TMV micrograph and for the successful run on my own micrograph. Let me know if I can provide more information to help continue the error search!

[Carsten Sachse]

Dear Collin,

We had another look at your parameter files and certified that they are running. It seems like there are some issues with installation. Are you using other python or EMAN2 versions than the ones provided by Spring? Are there other things that are non-standard installations?

Best wishes,

Carsten

PS For your convenience I also attach the parameter files that did clearly work in a number of our test environments.