SqueezeMeta 1.65post1 gives checkM errors and reinstall can't connect to host

[dutchscientist]

One of my virtual computers (Ubuntu 22.04 LTS) gives a problem with SqueezeMeta 1.65post1. It crashes at step 17, where the syslog gives:

[22 seconds]: STEP17 -> 17.checkM_batch.pl
PATH=/home/vetschool/miniconda3/envs/SqueezeMeta/SqueezeMeta/bin:/home/vetschool/miniconda3/envs/SqueezeMeta/SqueezeMeta/bin/pplacer:/home/vetschool/miniconda3/envs/SqueezeMeta/SqueezeMeta/bin/hmmer:$PATH /home/vetschool/miniconda3/envs/SqueezeMeta/SqueezeMeta/bin/checkm taxon_set genus Prevotella /home/vetschool/data/nisha/mg01_oral_deacon/data/checkm_markers/Prevotella.ms >> /home/vetschool/data/nisha/mg01_oral_deacon/syslog 2>&1
It seems that the CheckM data folder has not been set yet or has been removed. Running: 'checkm data setRoot'.
Where should CheckM store it's data?
Please specify a location or type 'abort' to stop trying:

Another virtual computer installed at the same time runs fine. Same database.

I have removed the conda environment, installed it again but then refuses to recognise the host (?) when running configure_nodb.pl (crashes on get_host.pl which can't reach a host). Rebooted, no difference. I even copied the SqueezeMeta folder from the running computer in my miniconda envs folder, crashes as well.

So not clear where it goes wrong?

Edit: have copied the whole folder to the virtual computer running fine, and it is progressing there. So it is the specific install that is dysfunctional, it is not the dataset used.

[fpusan]

One of the problems is that our server is down. So that's why configure_nodb.pl (which was the right call, as it seems it hadn't been set up before) is failing for you

[fpusan]

Server is back up

[dutchscientist]

Great, thank you!

[dutchscientist]

@fpusan I still get the same error message:
"No host could be reached! at /home/vetschool/miniconda3/envs/SqueezeMeta/SqueezeMeta/lib/install_utils/get_host.pl line 11."

[dutchscientist]

Is there anything else we can do to get configure_nodb.pl to work?

[jtamames]

Hello!
I had to stop the server today because of maintenance. It will be available again at the end of the week
Sorry for the inconvenience
Best,
J

[dutchscientist]

To come back to this one, the server was up and test_install.pl gives the all-clear.

But the checkM error keeps on returning on that virtual computer. Have already removed the conda environment completely and reinstalled (1.6.5.post1), am doing that again now and downloading the database. Have also cleaned up the Conda environment and made sure with conda remove to use the --all switch so to remove any settings related to the SqueezeMeta environment.

Any thoughts on how this can be resolved?

[dutchscientist]

Well, I downloaded the database again, and then it works fine. Put back the old database, fail. Completely baffling. The old one passes all the tests, except for the important one...

The old one has the nr.dmnd dated at 02/09/2023 (size 311,593,167 kb), the new one has as date 07/06/2024 (size 362,769,818 kb). As we have to make sure the databases match, I guess I need to put the newer one on all my virtual computers.

[fpusan]

Weird. The old database has definitely been working well in the past.
Did you run configure_nodb.pl to set it up after our server got back up?

[dutchscientist]

Yes, and all reported well. I have now just moved to the new database, will test the stations with the interrupted run to see checkM is working, and don't look back ;)

[fpusan]

test_install,pl should check for the checkm database being properly configured

SqueezeMeta/utils/install_utils/test_install.pl

Line 165 in 3a66a99

open( infile_, "$installpath/lib/checkm/DATA_CONFIG" ) || die ("\nCRITICAL ERROR: Can not find the checkm DATA_CONFIG file in $installpath/lib/checkm. This indicates a broken installation.\n\t\tIf the error persists after reinstalling from scratch please open an issue at http://github.com/jtamames/SqueezeMeta\n\n");

I assume going to the latest SqueezeMeta version is not an option, right? There we switch to CheckM2, which solves some bugs that you may run into with 1.6.5.

[dutchscientist]

Not at the moment, this is not for myself but for other people who need the consistency with the previous observations as much as possible.