CTFFind3/ctfestimate.py issues
Added by Anand Gupta over 12 years ago
We've been using CTFFind3 via appion to do CTF estimation, but it seems that recently ctfestimate.py changed and now includes a dependency on matplotlib. When we installed matplotlib, however, ctfestimate.py fails because it seems that there is an incompatibility issue between the current version of matplotlib (0.99.1.2) and the version of numpy we've been using (1.0.1) in CentOS. Specifically, matplotlib fails to import because it cannot import numpy.ma:
Traceback (most recent call last):
File "<stdin>", line 1, in ?
File "/usr/lib64/python2.4/site-packages/matplotlib/__init__.py", line 129, in ?
from rcsetup import defaultParams, validate_backend, validate_toolbar
File "/usr/lib64/python2.4/site-packages/matplotlib/rcsetup.py", line 19, in ?
from matplotlib.colors import is_color_like
File "/usr/lib64/python2.4/site-packages/matplotlib/colors.py", line 54, in ?
import matplotlib.cbook as cbook
File "/usr/lib64/python2.4/site-packages/matplotlib/cbook.py", line 10, in ?
import numpy.ma as ma
ImportError: No module named ma
It's rather difficult/generally infeasible to downgrade matplotlib, but might be possible to update numpy. That said we are wary of upgrading numpy because there are so many other dependents on it. We're wondering what versions of matplotlib and numpy these tools are being built with, as they are clearly different from the original ones listed on the wiki...
Replies (1)
RE: CTFFind3/ctfestimate.py issues - Added by Anchi Cheng over 12 years ago
Thank you for pointing out that numpy 1.0.1 is not compatible with current version of matplotlib. We will have to update the minimal version for numpy when we release future branches that uses matplotlib.
It looks like you have been using the trunk of the svn. It is under active development. At moment, there is a fair amount of development regarding ctf estimation and correction and can be very unstable and there are still fine-tuning effort based on alph tester/developer's feedback. Documentation and support can not keep up with this kind of daily changes. I would recommend that you use the branch 2.2 for now. It was branched off and kept up to date in term of debugging but not merged with fresh new development like the change you indicate.
That said, we are currently running numpy 1.2.1 and 0.98.5.2 matplotlib on most of our processing servers at NRAMM and we have not had issue with other Leginon/Appion functions.
Anchi