Closed Amy1809 closed 1 month ago
Looks like a permissions issue ('/bin/sh: 1: cryoCARE_train.py: Permission denied' being the clue). The 'could not find denoise model' error message usually just means cryoCARE has failed to run. Try running cryoCARE outside of RELION and see what happens:
cryoCARE_train.py --conf Denoise/jobXYZ/external/train_config.json
(or wherever the train_config.json is)
If it still won't work, it's an issue with your cryoCARE installation/permissions (what I suspect)
Looks like a permissions issue ('/bin/sh: 1: cryoCARE_train.py: Permission denied' being the clue). The 'could not find denoise model' error message usually just means cryoCARE has failed to run. Try running cryoCARE outside of RELION and see what happens:
cryoCARE_train.py --conf Denoise/jobXYZ/external/train_config.json
(or wherever the train_config.json is)
If it still won't work, it's an issue with your cryoCARE installation/permissions (what I suspect)
Thank you for your reply. I have tried running cryoCARE outside RELION and it worked. I carried on using this trained model to run CryoCARE predict in RELION, which worked but I noticed something in the output file given by RELION:
[07:56:44] Generating denoised tomograms predict.py:108
2024-07-11 07:56:47.623123: W tensorflow/stream_executor/gpu/asm_compiler.cc:63] Running ptxas --version returned 256
2024-07-11 07:56:47.696499: W tensorflow/stream_executor/gpu/redzone_allocator.cc:314] Internal: ptxas exited with non-zero error code 256, output:
Relying on driver to perform ptx compilation.
Modify $PATH to customize ptxas location.
This message will be only logged once.
Looking for GPU with ID: 0
GPU 0 successfully found
Loading network weights from 'weights_best.h5'.
(168, 648, 648, 1)
100%|██████████| 8/8 [00:08<00:00, 1.05s/it]
['' '' '' '' '' '' '' ''
'cryoCARE 11-Jul-24 07:57:12 '
'']
Loading network weights from 'weights_best.h5'.
(168, 648, 648, 1)
100%|██████████| 8/8 [00:08<00:00, 1.04s/it]
['' '' '' '' '' '' '' ''
'cryoCARE 11-Jul-24 07:57:22 '
'']
Loading network weights from 'weights_best.h5'.
(168, 648, 648, 1)
100%|██████████| 8/8 [00:08<00:00, 1.03s/it]
['' '' '' '' '' '' '' ''
'cryoCARE 11-Jul-24 07:57:32 '
'']
Loading network weights from 'weights_best.h5'.
(168, 648, 648, 1)
100%|██████████| 8/8 [00:08<00:00, 1.04s/it]
['' '' '' '' '' '' '' ''
'cryoCARE 11-Jul-24 07:57:43 '
'']
Loading network weights from 'weights_best.h5'.
(168, 648, 648, 1)
100%|██████████| 8/8 [00:08<00:00, 1.05s/it]
['' '' '' '' '' '' '' ''
'cryoCARE 11-Jul-24 07:57:53 '
'']
[07:57:54] Denoised tomograms successfully generated, finalising predict.py:116
metadata
There is not error message. By looking at the beginning of the output file, it says a non-zero error code 256, which I don't know what it means. I used this denoised tomograms for particle picking, the contrast has improved but was not very good.
good that it works now. Don't worry about that error message. Tensorflow tends to throw out loads of that type of message always, as long as it has worked it should be good.
As an alternative, maybe try IsoNet, I probably like this denoising software more than CryoCARE
good that it works now. Don't worry about that error message. Tensorflow tends to throw out loads of that type of message always, as long as it has worked it should be good.
As an alternative, maybe try IsoNet, I probably like this denoising software more than CryoCARE
I will try that, thanks so much on the help! :)
I am following the Relion 5 pipeline starting from the very beginning. When I proceed to the frist step (training setp) in denoising, Relion5 gives me an error saying that the it could not find the training model in my denoise job output. I have cryoCARE installed, and I have set 'Generate tomograms for denoising' to Yes in both motion correction and tomogram reconstruction stages, so I don't know why this happens.
Environment:
Job options:
which relion_python_tomo_denoise
cryoCARE:train --tomogram-star-file Tomograms/job038/tomograms.star --output-directory Denoise/job039/ --gpu 0 --training-tomograms Position_1:Position_1_2:Position_2_2:Position_3_3:Position_6 --number-training-subvolumes 1200 --subvolume-sidelength 72 --pipeline_control Denoise/job039/Error message:
And here is the .out file
Any suggestions or advice will be helpful.