For the past few days I have been trying to use Cufflinks for my RNA seq analysis but I have been repeatedly getting this error:
slurmstepd: error: couldn’t chdir to `/srv/pulsar/main/venv/lib/python3.6/site-packages’: No such file or directory: going to /tmp instead
This started happening a few days ago and up until then I had no issues running the tool at all. My inputs are BAM file obtained from HISAT2 alignment and gtf file for mm10 reference annotation. I noticed that the assembled transcripts gtf file still outputted from Cufflinks despite the error, so I tried inputting these assembled transcripts into Cuffmerge (along with some successfully obtained assembled transcript gtf’s before this error started recurring), but Cuffmerge also resulted in an error:
[Fri Aug 28 23:06:48 2020] Preparing output location output/
Traceback (most recent call last):
File “/cvmfs/main.galaxyproject.org/deps/_conda/envs/__cufflinks@2.2.1/bin/cuffmerge”, line 580, in
sys.exit(main())
File “/cvmfs/main.galaxyproject.org/deps/_conda/envs/__cufflinks@2.2.1/bin/cuffmerge”, line 538, in main
gtf_input_files = test_input_files(transfrag_list_file)
File “/cvmfs/main.galaxyproject.org/deps/_conda/envs/__cufflinks@2.2.1/bin/cuffmerge”, line 268, in test_input_files
g = open(line,“r”)
IOError: [Errno 2] No such file or directory: ‘input_1 input_2 input_3 input_4 input_5’
I have been able to successfully use this tool without any problems up until now. Any solutions would be appreciated; I’m still relatively new to Galaxy.
From their error: slurmstepd: error: couldn’t chdir to `/srv/pulsar/main/venv/lib/python3.6/site-packages’: No such file or directory: going to /tmp instead
I found that people outside of Galaxy Project can have problems with slurmstepd, so there’s probably something wrong inside GP rather than with your jobs
I’ll write them a report and see if they know what’s happening.
The issue with Cufflinks appears to be similar but that is not confirmed. Cufflinks is a deprecated tool and is no longer supported. However, you can certainly try the same options below and see what results.
Note: This part of the error message is just a log message about rerouted resources and isn’t related to the bug uncovered related to Pulsar/Jetstream/Stringtie. Even successful jobs can report it in a log message. These two tools, plus others.
slurmstepd: error: couldn’t chdir to `/srv/pulsar/main/venv/lib/python3.6/site-packages’: No such file or directory: going to /tmp instead
For Stringtie, our administrator @nate will be re-routing all jobs for this tool to the cluster where it works while resolving the root bug. Not sure that is done or not yet (testing in progress), but you can try two ways to get the tool run successfully at the cluster where it works right now – you don’t need to wait for us to fix it the problem or to confirm the temporary configuration change. You can try both ways at the same time to get this done quicker (and purge one of the results to recover disk space should both work).
Rerun again using defaults for the “Job Resource Parameters”. This is probably what you are doing now – your two matched jobs were sent to different clusters, and one is problematic for this tool (Jetstream). If this is successful, then you can keep using this method. Is simpler, and once this bug is fixed, you’ll want Galaxy to choose the appropriate cluster for most use cases (this tool and others). Jobs are scheduled to clusters based on data/tool factors and cluster node availability at any particular time.
Rerun again, choosing the cluster to send the job to. Under “Job Resource Parameters” set the option to “Galaxy cluster (default)”. The problem is at the “Jetstream” cluster – and this action specifically avoids it.
