bug in running tutorial

[ajian-he]

I install v0.2.3 when I run the tutorial, as well as my own data, it always come to the error like below: UserWarning: Couldn't retrieve source code for container of type LinearClassifier. It won't be checked for correctness upon loading. UserWarning: Couldn't retrieve source code for container of type ResNet8. It won't be checked for correctness upon loading. UserWarning: Couldn't retrieve source code for container of type Sequential. It won't be checked for correctness upon loading. UserWarning: Couldn't retrieve source code for container of type BasicConv2d. It won't be checked for correctness upon loading. UserWarning: Couldn't retrieve source code for container of type Conv2d. It won't be checked for correctness upon loading. UserWarning: Couldn't retrieve source code for container of type ReLU. It won't be checked for correctness upon loading. UserWarning: Couldn't retrieve source code for container of type ResidA. It won't be checked for correctness upon loading. and the program stuck! please help! Thanks!

[tbepler]

This sounds like a problem with your install path. How did you install topaz? What system are you on?

[ajian-he]

Thanks. I follow the instruction, first I install Anaconda, then create a conda environment for Topaz as suggested "conda create -n topaz python=3.6...", finally I install topaz via "conda install topaz -c tbepler -c pytorch" . The system is CentOS 7.6. Really need help, thanks!

[tbepler]

What version of pytorch do you have installed? You can find this if you activate your conda topaz environment and run conda list:

source activate topaz
conda list

[ajian-he]

here is the outcome, thanks!

packages in environment at /share/XXX/anaconda3/envs/topaz:

#

Name Version Build Channel

_libgcc_mutex 0.1 main ca-certificates 2019.11.27 0 certifi 2019.11.28 py37_0 libedit 3.1.20181209 hc058e9b_0 libffi 3.2.1 hd88cf55_4 libgcc-ng 9.1.0 hdf63c60_0 libstdcxx-ng 9.1.0 hdf63c60_0 ncurses 6.1 he6710b0_1 openssl 1.1.1d h7b6447c_3 pip 19.3.1 py37_0 python 3.7.5 h0371630_0 readline 7.0 h7b6447c_5 setuptools 42.0.2 py37_0 sqlite 3.30.1 h7b6447c_0 tk 8.6.8 hbc83047_0 wheel 0.33.6 py37_0 xz 5.2.4 h14c3975_4 zlib 1.2.11 h7b6447c_3

[tbepler]

It looks like topaz isn't installed into your topaz conda environment. Try running 'source activate topaz' and then 'conda install ...' again. That should fix the problem.

[ajian-he]

I tried to de novo install all conda and topaz, follow the suggestion of first activate topaz and then conda install..., the issue still exist, anyway, I can continue all tutorial, but the outcome is slightly different from the demo tutorial

[tbepler]

There is some randomness in the results due to how the neural network parameters are optimized. Are you getting the same warning, but the everything otherwise runs normally?

[ajian-he]

the training results are different from the tutorial: epoch iter split loss ge_penalty precision tpr fpr auprc 1000 1 1001 test 0.015258 - 0.040877 0.258628 0.009914 0.05945546523116192 2001 2 2001 test 0.015993 - 0.044977 0.294445 0.010214 0.061771123509967794 3002 3 3001 test 0.020187 - 0.042730 0.374459 0.013705 0.06958290330218277 4003 4 4001 test 0.020249 - 0.045742 0.388267 0.013233 0.0675878506697664 5004 5 5001 test 0.021257 - 0.047583 0.420393 0.013747 0.0704535301988471 6005 6 6001 test 0.022189 - 0.040423 0.400547 0.015534 0.07013125083592019 7006 7 7001 test 0.024289 - 0.050577 0.467966 0.014352 0.07362886436482818 8007 8 8001 test 0.019194 - 0.055014 0.402074 0.011283 0.07455601152215174 9008 9 9001 test 0.018753 - 0.049260 0.378237 0.011926 0.07213143960590812 10009 10 10001 test 0.017719 - 0.058193 0.401393 0.010613 0.0823831178371298

[ajian-he]

and the picked particles is much less than tutorial

[tbepler]

As I said, there is randomness from the optimizer, so every run won't get exactly the same results. These results look normal. Just lower the score threshold to get more particles. In the tutorial, this threshold is set in the "topaz particle_stack" and micrograph visualization steps.

[ajian-he]

thanks, I am running my own data now, let's see how it lood

[ajian-he]

tbepler, I found that topaz indeed found many previous hided or unfound particles, however, I have issues either use the coordinates from topaz to extract particles in relion, or use the topaz extracted particles to do further analysis (e.g. class 2d, 3d) in relion as they lack ctf or defocus values. I try to use the relion micrographs_ctf.star as metadata to combine the data, but failed. Any suggestion, please. Thank you.

[tbepler]

What do you mean by it failed? Did you get an error from "topaz particle_stack" when you tried to include your metadata?

I believe you can load the particles directly into relion (using the .star file from extract) and add CTF information there. @alexjnoble can you chime in on adding CTF information in relion?

[ajian-he]

when I use import job in relion to import the topaz particle256.star, the import job seems ok, but when I later use the import to do 2D classification, it shows error like this: ERROR: Cannot read file particles256.mrc It does not exist [localhost.localdomain:101565] 3 more processes have sent help message help-mpi-api.txt / mpi-abort [localhost.localdomain:101565] Set MCA parameter "orte_base_help_aggregate" to 0 to see all help / error messages

when I tried to directly load particle256.star into relion 2D classification, it also shows : ERROR: Cannot read file particles256.mrc It does not exist

Could you give me a step-by-step guide how to load the topaz extract to relion, either the particle.star file way, or the coordinate file way? Thanks a lot!

[ajian-he]

Figured out, use re-extract, thanks!

[tbepler]

Great, I'm glad to hear it's working. I'm going to close this issue since it sounds like everything is resolved.