Appion

​Do you have Gctf set up so that it uses the frames for each image as input? Gctf is able to tile from each frame of an image if you feed it a frame stack (.mrcs format) for each image.

For your setup, do you have a GPU connected directly to your Appion server? How are you actually running it?

The current Appion Gctf implementation doesn't use the frames from each image as input, but I will add this in.

We have four GPU nodes in our cluster where we run Gctf jobs. I made some changes to TorqueHost.py, processingHost.py, and added an apGpuJob jobtype. The only significant difference is that the .job headers say "nodes=1:gpus=1" instead of "nodes=1:ppn=x". Show command and copy/paste to your GPU machine should work if you don't have GPU nodes. Also the executable in appion/bin/gctf.py should be changed to match the version you are using. Thanks

Hey Carl,

Scott and I are not seeing Gctf show up on the single particle Estimate the CTF/CTF Estimation Procedures section of the Appion single particle processing webpage. From your revision history it looks like you only created/edited gctf.inc and gctf.py. Have you not committed the php code that links your gctf.inc webpage?

-Alex

I added the following flag to myamiweb/config.php.template so people running appion without GPU support can prevent GPU programs from displaying:

define('HIDE_GPU_FEATURE',false);

Add this to myamiweb/config.php and it should become visible. Thanks

That's a good idea! Is this documented somewhere so others will be aware? I'll be adding some GPU software this year and I don't want it to be default hidden... maybe it would be better to make it default visible, but tell you repeatedly that you need to have a GPU to run this code?