RCT problems on myami-2.2-trunk
Added by Michael Cianfrocco over 10 years ago
We have two versions of appion running in our lab, a stable version (2.2 branch) and the trunk version for 2.2 (Revision 17931). Unfortunately, it seems that the euler angles from XMIPP ML2D are incorrectly used during the RCT reconstruct step in the trunk version (given that the resulting volume has severe artifacts indicative of incorrect euler angle conversion). This is contrast the stable version, which is capable of generating correct RCT reconstructs from XMIPP ML2D.
Has anyone else seen this? As far as I can tell, rctVolume.py, maxlikeAlignment.py and the associated libraries within apTilt/ appear to be the same. I'm uploading RCT tilt pairs using the image uploader from the same dataset that works in the stable branch of Appion.
I've repeated the error a few times to convince myself that it isn't the way that I'm picking particles or creating stacks, but that it is a problem within appion. However, it still could be me (somehow).
Thank you for your help,
mike
Replies (8)
RE: RCT problems on myami-2.2-trunk - Added by Amber Herold over 10 years ago
Hi Mike,
I can take a look into this tomorrow to see what the issue may be. I'll get back to you then.
Amber
RE: RCT problems on myami-2.2-trunk - Added by Michael Cianfrocco over 10 years ago
Hi Amber,
Thank you for looking into this. I wanted to follow up by saying that I just repeated this error with a different dataset at a negative tilt angle (instead of a positive tilt which was used in the above example). For each case, the theta tilt angle is correct within the euler angle file, which makes me think it is the in-plane rotation angle conversion that is wrong or particle tilt-mate selection. But, considering that the euler angle conversion appears to be the same for the 2.2-branch vs. the 2.2-trunk, I'm wondering if it is a bug in the selection of tilt-mates.
To test if that was the problem, if I were to create a particle stack for all the untilted particles that went into the class average, should it be the same particle mates that are extracted for the RCT reconstruction? I don't know off the top of my head, so I'll look into this in the meantime.
Thank you,
Mike
RE: RCT problems on myami-2.2-trunk - Added by Amber Herold over 10 years ago
Hi Mike,
I'm in the process of trying to reproduce this. It is busy around here this week and next week, so it is a bit slow going. I did notice that there is later revision for 2.2, r17938, however I don't believe it should make a difference.
Amber
RE: RCT problems on myami-2.2-trunk - Added by Michael Cianfrocco over 10 years ago
Hi Amber,
I'm sure things have been busy, but have you been able to reproduce any of my problems? Or is this someone else you know who is using / has used the trunk version of appion recently for RCT?
Thank you for your help,
Mike
RE: RCT problems on myami-2.2-trunk - Added by Amber Herold over 10 years ago
Hi Mike,
I put out a message to other users and developers last week, and it seems that none have experienced this particular issue.
One thing I should have asked you earlier - What do you mean when you say 2.2 branch vs 2.2 trunk?
I had assumed that you meant that you installed a stable version with a tar file and another with svn that you update with svn. Or possibly that you had an older and a newer revision of the same branch.
We have one 2.2 branch. The the svn trunk is for current development. Any time we put out a new version, we create a branch for that version. Any updates to that version are added to it's branch. Can you tell me which revision your stable working version is at?
Have you checked to see if any database updates were needed between the older and the newer revision? You can do the following to see if there are any update scripts that need to be run for the newer version to work.
cd /your_download_area/myami/dbschema python schema_update.py
Is there any chance that you may be ready to upgrade to the 3.0 version? It requires upgrading from CentOS 5.x to CentOS 6.x, so it is not a small update. However, we have not used 2.2 for nearly 2 years now, so supporting it is becoming more and more difficult.
Thanks for your patience on this the last couple of weeks.
Best,
Amber
RE: RCT problems on myami-2.2-trunk - Added by Amber Herold over 10 years ago
Can you send 1 pair of micrographs that can be used to try to reproduce this?
Thanks
RE: RCT problems on myami-2.2-trunk - Added by Michael Cianfrocco over 10 years ago
Hi Amber,
Thank you for your patience! It turns out that our database needed updating - now the RCT works! I guess I'm still learning how and when to update the svn repository and the database.
And you're also correct that I'm running a branch of myami-2.2-redux. For some reason I got confused and though it was the trunk.
We'd love to transition to v3.0, but we are slow to change here, we just got redux up and running. After updating our machines from CentOS 5 to Scientific linux 6.5
As a side queustion - is CentOS 6 comparable to Scientific Linux 6.5?
Thank you again for your help in clearing up my confusion!
mike
RE: RCT problems on myami-2.2-trunk - Added by Amber Herold over 10 years ago
Aha!
Yes, running 2.2 and 2.2 redux are not compatible with each other. The way we define ctf values internally has changed between those versions. The database updates that you ran have changed all your existing db values to the new standard. For this reason it is important that you do NOT run the older 2.2 branch again. Your ctf values will be incorrect if you do.
Since you have already moved to 2.2-redux, moving to 3.0 will be much simpler when you are ready.
I believe scientific linux 6.5 should be fine, just make sure you have at least python 2.6 and php 5.3.
Now, for when to run the db update script. Every time that you svn update, you will want to run the script to make sure bad things don't happen.
Cheers,
Amber