ERROR: sequence ss_pred contains no residues.

[1000000000000000000000000000000]

Hello everyone,

I am having a difficult time using RoseTTAfold as I am inexperienced with using HPC and more complex installation procedures. I am running into a few problems while running the software.

Log directory contains 6 files:

hhsearch.stderr file:

The hhsearch.stdout file contains 0 lines.

make_mas.stderr

make_msa.stdout

make_ss.stderr

make_ss.stdout

The hhblits dir contains 3 *.a3m files

I appreciate any advice and suggestions to complete a successful installation. Also, when the program is installed successfully, optimizing the computing resources might be very helpful also because I have had a tough time allocating appropriate computing nodes to this program so far on the HPC.

Thanks all!!!!

[1000000000000000000000000000000]

FYI:

This is the GCC version I am working with:

[Meowooo]

I had the same problem

[Sabryr]

I faced the same issue and my solution was to remove hhsuite from conda and then compile my own. We use the EasyBuild module system and I am using HH-suite/3.3.0-foss-2020a module.

[1000000000000000000000000000000]

Is my NVIDIA driver or GCC version a possible problem? I installed using the "conda env create -f RoseTTAFold-linux-cu101.yml" command.

I think there might be an issue with my PSIPRED as opposed to my HHSUITE because I am seeing some strange file format issue.

The error I am referring to says, "Bad psipred pass1 file format!" Thoughts?

@Sabryr Would an incompatible HHSUITE cause this error? If so, how would I fix this? Also, thank you for the response!

[Sabryr]

I am using CUDA11 with NVIDIA A100, so can not say for sure if CUDA 10 is an issue. I did also hit the same error you reported "hhalignment.cpp 223" Then I removed hhsuite from conda and installed my own, now the specific error is gone.

[1000000000000000000000000000000]

Hmm, okay. I know how to remove hhsuite from the RoseTTAFold conda environment but how should I install hhsuite? Is there a better version somewhere? Does it need to be installed on the PATH? Does it need to be in a specific environment?

Thanks for the help

[Sabryr]

Version is exactly the same. I complied it from source code instead of the per-compiled binary in conda. Yes the path needed to be set correctly. For me this was easy as we already use the EasyBuild frame work on our system (https://github.com/easybuilders/easybuild-easyconfigs/blob/develop/easybuild/easyconfigs/h/HH-suite/HH-suite-3.3.0-foss-2020a.eb),

[1000000000000000000000000000000]

I uninstalled the hhsuite then compiled my own version and I seem to still get the same errors. I am wondering if I am doing this correctly.

[1000000000000000000000000000000]

Okay, I have decided to try and use easybuild to install the hhsuite. Here is how I attempted this:

1. I removed the RoseTTAFold conda environment
2. I edited the yml file for the RoseTTAFold environment to exclude hhsuite
3. I reinstalled RoseTTAFold conda env using this edited yml file.
4. In the terminal I ran srun --exclusive --time 1:0:0 --pty bash -i command to use compute node for the install.
5. I used conda activate RoseTTAFold to enter into the conda environment
6. I used conda install -c conda-forge easybuild command to attempt to install easybuild <---- THIS FAILED
7. I then used pip install easybuild command <---- THIS WORKED
8. When typing eb --version I got easy build version 4.4.1
9. I copied the text from the link you sent for the HH-Suite-3.3.0-foss-2020a.eb file and saved it
10. I used the following command to install hhsuite using easybuild: eb --modules-tool EnvironmentModulesC --module-syntax Tcl --robot HH-suite-3.3.0-foss-2020a.eb

This seems to be working but the installation is still ongoing. Does this seem like a reasonable approach so far?

Thanks!

[Sabryr]

If you have not used EasyBuild before then you might face more issue, not sure how the module system is setup on your system either. HH-Suite-3.3.0-foss-2020a.eb would also install the compilers etc.. so it will take some time yes. I mentioned EasyBuild as it was already setup on my system, sorry if I gave the impression that EasyBuild would solve your issue. If I were you I would investigated why the H-suite you compiled did not work first.

[1000000000000000000000000000000]

Oh, i see. Hmm, not sure what to do now.

@minkbaek Do you have any suggestions for me? I am having a tough time with this installation. I am determined to be successful though! :)

[Meowooo]

Have you solve this question? @1000000000000000000000000000000 I have been troubled in this problem for a long time

[1000000000000000000000000000000]

Unfortunately, I have not yet solved this issue. Easybuild did not work for me so I am going to uninstall and try to reset my installation to reflect the original install steps defined by their github repo. After that I will try to reassess what my environment's issue might be.

[ro6ert]

Did you have separate environments for rosetta and folding or did you install the folding.yml into the PyRosetta env?

[1000000000000000000000000000000]

I have separate conda environments.

Should I install folding into the pyrosetta env?

Thanks!!!

On Sat, Aug 28, 2021 at 8:29 AM Robert Chase @.***> wrote:

Did you have separate environments for rosetta and folding or did you install the folding.yml into the PyRosetta env?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/RosettaCommons/RoseTTAFold/issues/68#issuecomment-907642135, or unsubscribe https://github.com/notifications/unsubscribe-auth/ATPRT6CA52ROH7I7JPUXLB3T7D6D7ANCNFSM5CEGV5EQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

[ro6ert]

This line of the run_pyrosetta_ver.sh script sources the pyrosetta environment, not the folding environment.

https://github.com/RosettaCommons/RoseTTAFold/blob/main/run_pyrosetta_ver.sh#L30

The README says, "The latter is required to run a pyrosetta version only". I think this is a bug in the documentation because by default the installation will make a separate folding environment rather than adding to the pyrosetta environment.

If you use a default installation using conda of the folding.yml file it will be installed in the "folding environment"

https://github.com/RosettaCommons/RoseTTAFold/blob/main/folding-linux.yml#L1

You can install the folding-linux yaml file into pyrosetta enviroment by using conda install -n pyrosetta folding-linux.yml

I'm having trouble getting pdb files out of the run_pyrosetta_ver.sh and I got the error message you posted on, so this may be related to this other issue https://github.com/RosettaCommons/RoseTTAFold/issues/50. In the other thread, they suggest that the environment installation process may be to blame. Their solution is to use something called easy-install.

I would like to try to see if the end to end script runs as a first debugging experiment and then see if installing the linux-folding yaml into the pyrosetta environment using conda install will work as a second experiment, but it will be another 17 hours for the code to finish running if it runs like yesterday's attempt.

[Sabryr]

I would just like to share the following info. When I was trying to install RoseTTAFold on another system, with the same setup I again encountered this error. Although I had used the same HH-suite/3.3.0-foss-2020a version. Then during comparison I found that i had an undocumented change I did on the first system. which was to set BLASTMAT. More detail here https://github.com/Sabryr/ProteinFolding/issues/3 https://github.com/Sabryr/ProteinFolding/issues/2

[1000000000000000000000000000000]

@ro6ert You are right that the run_pyrosetta_ver.sh script activates the RoseTTAFold conda environment on line 30. I am not suspecting this to be the problem only because on line 84 that conda env is deactivated and on line 85 the conda env folding is activated. Any thoughts on this? If this worked for you though let me know.

[1000000000000000000000000000000]

@Sabryr

I am not really sure how I would modify the following Path in the line you added to the /RoseTTAFold/input_prep/make_ss.sh script:

export BLASTMAT=/cluster/software/RoseTTAFold/1.0.0.2/share/blast-2.2.26/data

I do not have this path. I do have a /RoseTTAFold/csblast-2.2.3/data path though. Do you think it could work if I used that path instead?

Thanks!!

[Sabryr]

This will be in your conda environment. You could do a search inside that if you know where the conda environment is. eg. 1.cd <PATH to CONDA Install>/ 2. find . -name blast-2.2.26. If the blas version is different then you could look for the folder named shared (find . -name share) If you do not know where exactly the conda environment is setup. Then you could use which python or which pip while the environment is active, to get the location.

[1000000000000000000000000000000]

Hmm, there does not seem to be any blast folder in share. See below:

There are some programs with blast in the name in the bin directory however. See here:

[Sabryr]

Then some dependencies may have not been installed properly. Please cross check with the procedure in RoseTTAFold provided and also somewhat customized version I had used https://github.com/Sabryr/ProteinFolding/tree/master/RoseTTAFold/install

[275145]

Unfortunately, I have not yet solved this issue. Easybuild did not work for me so I am going to uninstall and try to reset my installation to reflect the original install steps defined by their github repo. After that I will try to reassess what my environment's issue might be.

Hello, have you solved this problem now, I have been troubled by it for a long time

[275145]

Have you solve this question? @1000000000000000000000000000000 I have been troubled in this problem for a long time

Have you solved this problem, I also encountered the same problem, I have been troubled by this problem for a long time @Meowooo

[1000000000000000000000000000000]

@275145 The short answer is no but I think it has to do with HHBlitz not running. I am wondering if there is a HHBlitz path variable that I need to change somewhere in a config file or something. I think your strategy about reinstalling everything is reasonable and I am probably going to do the same thing, especially since there has been recent github comits.