The bioBakery help forum

Error message returned from diamond


when trying to run HUMAnN3 I retrieve an error message returned from diamond:

Error message returned from diamond :
diamond v0.9.24.125 | by Benjamin Buchfink
Licensed under the GNU GPL
Check for updates.

#CPU threads: 4
Scoring parameters: (Matrix=BLOSUM62 Lambda=0.267 K=0.041 Penalties=11/1)
Temporary directory: Text file busy
Error: Error calling unlink.

06/29/2020 03:37:26 PM - humann.utilities - CRITICAL: TRACEBACK:
Traceback (most recent call last):
File “/home/plicht/anaconda3/envs/metaphlan/lib/python3.7/site-packages/humann/”, line 744, in execute_command
p_out = subprocess.check_output(cmd, stderr=subprocess.STDOUT)
File “/home/plicht/anaconda3/envs/metaphlan/lib/python3.7/”, line 411, in check_output
File “/home/plicht/anaconda3/envs/metaphlan/lib/python3.7/”, line 512, in run
output=stdout, stderr=stderr)
subprocess.CalledProcessError: Command ‘[’/home/plicht/anaconda3/envs/metaphlan/bin/diamond’, ‘blastx’, ‘–query’, ‘/media/sf_projects/microbiome/Analysis_of_microbiome/WiP/KneadData/firsttry/HUMAnN3/output/TRIAL_PL018_1_Novogenea1_1/TRIAL_PL018_1_Novogenea1_1_humann_temp/TRIAL_PL018_1_Novogenea1_1_bowtie2_unaligned.fa’, ‘–evalue’, ‘1.0’, ‘–threads’, ‘4’, ‘–top’, ‘1’, ‘–outfmt’, ‘6’, ‘–db’, ‘/media/sf_projects/microbiome/Analysis_of_microbiome/BioBakery-Tools/Databases/HUMAnN_db/uniref/uniref90_201901’, ‘–out’, ‘/media/sf_projects/microbiome/Analysis_of_microbiome/WiP/KneadData/firsttry/HUMAnN3/output/TRIAL_PL018_1_Novogenea1_1/TRIAL_PL018_1_Novogenea1_1_humann_temp/tmpln4cfsa3/diamond_m8_18tgcgi4’, ‘–tmpdir’, ‘/media/sf_projects/microbiome/Analysis_of_microbiome/WiP/KneadData/firsttry/HUMAnN3/output/TRIAL_PL018_1_Novogenea1_1/TRIAL_PL018_1_Novogenea1_1_humann_temp/tmpln4cfsa3’]’ returned non-zero exit status 1.

I use HUMAnN3 alpha3 with Diamond v0.9.24.125 installed via conda, and as databases:
-uniref full ($ humann_databases --download uniref uniref90_diamond )

  • ChocoPhlAn full v296_201901 ($ humann_databases –download chocophlan full )
  • Uility mapping full ($ humann_databases --download utility_mapping full )

The functional tests with $ humann_test --run-functional-tests-tools --run-functional-tests-end-to-end are fine. Also when running the demo.fastq with the demo databases HUMAnN3 is working properly. Can you help me out?

Sorry for the delay here. Everything about the installation seems fine. This might be an issue with the I/O in your computer environment (per the “file busy” error). You could try re-running, or using another location to store the outputs?

Hi Eric,

thanks for your suggestions. Indeed, I guess the I/O directory used could be a problem since I am currently running a virtual ubuntu machine with outrunning disk space. Thtas why I located the databases as well as I/O files on an external HDD.
However, in the meantime I was able to get the task running with an updated diamond from v0.9.24.125 to v0.9.36.137. The command works well through with full Chocophlan_db and demo protein_db uniref90_demo_prots_v201901. However, when using the full uniref90_db, the command dies with <Signals.SIGKILL: 9>. I googled and watched my RAM and I guess it’s caused by running out of memory. The virtual machine is allocated with 8 GB RAM. Is there a general formula of minumum requirements of HUMAnN3?

In our evaluations on metagenomes with 30M reads, peak RAM usage varied from 16GB to 24GB depending on the balance between nucleotide and translated search. The memory ceiling is typically smaller outside of translated search, though this can also depend on the complexity of the microbial community under study.