sqm_reads.pl error at diamond stage

[bentsch]

Hi guys,

I recently installed the latest version of SQM via conda (v.1.7.2) including the databases.

I am trying to incorporate the sqm_reads.pl script into one of my workflows. However, I can not get it to work and wanted to ask if you maybe had an idea how to solve this.

I tried using the script on some of my own reads and also on the mock reads that come with the sqm database, I am getting the same error in both instances.

sqm_reads.pl -p mock_sqmreads -s mock_samples_file -f /scratch/c7183364/databases/sqm_202602/test/raw -t 2

SqueezeMeta on Reads v1.7.2, June 2025 - (c) J. Tamames, F. Puente-Sánchez CNB-CSIC, Madrid, SPAIN

This is part of the SqueezeMeta distribution (https://github.com/jtamames/SqueezeMeta)
Please cite: Tamames & Puente-Sanchez, Frontiers in Microbiology 10.3389 (2019). doi: https://doi.org/10.3389/fmicb.2018.03349

  Setting block size for Diamond
  AVAILABLE (free) RAM memory: 220.94 Gb
  We will set Diamond block size to 16 (Gb RAM/8, Max 16).
  You can override this setting using the -b option when starting the project, or changing
  the $blocksize variable in SqueezeMeta_conf.pl
Now reading samples from mock_samples_file
1 metagenomes found

SAMPLE 1/1: sample1

   File: mock.500k.S1_R1.fastq.gz
[1 seconds]: Running Diamond (Buchfink et al 2015, Nat Methods 12, 59-60) for taxa (GenBank nr, Clark et al 2016, Nucleic Acids Res 44, D67-D72)
Error running command /gpfs/gpfs1/scratch/c7183364/.conda/envs/squeezemeta_1.7/SqueezeMeta/bin/diamond blastx -q /scratch/c7183364/databases/sqm_202602/test/raw/mock.500k.S1_R1.fastq.gz -p 2 -d /gpfs/gpfs1/scratch/c7183364/databases/sqm_202602/db/nr.dmnd -e 0.001 --quiet -f tab -b 16 -o mock_sqmreads/sample1/mock.500k.S1_R1.fastq.gz.tax.m8 at /scratch/c7183364/.conda/envs/squeezemeta_1.7/bin/sqm_reads.pl line 248.

here is a look at the files it created during the run:

la
total 4
drwxr-xr-x 3 c7183364 c718 4096 Mar 17 10:27 .
drwxr-xr-x 7 c7183364 c718 4096 Mar 17 10:27 ..
-rw-r--r-- 1 c7183364 c718   28 Mar 17 10:27 creator.txt
-rw-r--r-- 1 c7183364 c718  148 Mar 17 10:27 methods.txt
-rw-r--r-- 1 c7183364 c718  165 Mar 17 10:27 mock_sqmreads.out.allreads
-rw-r--r-- 1 c7183364 c718  181 Mar 17 10:27 mock_sqmreads.out.mappingstat
drwxr-xr-x 2 c7183364 c718 4096 Mar 17 10:27 sample1
bash: __vte_prompt_command: command not found...
(squeezemeta_1.7) (JobID 5956827)[c7183364@n007.intern.leo5 mock_sqmreads]$ la sample1/
total 1
drwxr-xr-x 2 c7183364 c718 4096 Mar 17 10:27 .
drwxr-xr-x 3 c7183364 c718 4096 Mar 17 10:27 ..
-rw-r--r-- 1 c7183364 c718    0 Mar 17 10:27 mock.500k.S1_R1.fastq.gz.tax.m8

Since I also have sqm version 1.4 installed on my system, I also tried running sqm_reads.pl with that version. Interstingly, using that version, diamond seems to run successfully, but then I get an error at the next step (lca).
here is the last part of what was printed on the terminal:

[1 seconds]: Running Diamond (Buchfink et al 2015, Nat Methods 12, 59-60) for taxa (GenBank nr, Clark et al 2016, Nucleic Acids Res 44, D67-D72)
[12 seconds]: Running LCA
  Splitting Diamond file
  Starting multithread LCA in 2 threads: 1 2 Died at /scratch/c7183364/.conda/envs/squeezemeta_1.4.0/bin/sqm_reads.pl line 253.

after installing sqm v.1.7.2 I also did the test run using the mock reads and everything finished without an error; thus it seems that in general diamond is working in my environment. so maybe this is specifically related to the sqm_reads.pl script?

Any help will be highly appreciated.

Thanks a lot and best wishes

[fpusan]

Is this running on a cluster?
We try to set up the memory consumed by DIAMOND automatically, by looking at the free memory. The problem is that this memory may not be actually available to you (e.g. when running on a shared node).

[bentsch]

Hi,

thanks for your quick reply.

Yes, I am running this on a cluster. since I am currently still testing things, I run jobs within an interactive session. I just checked and for this interactive session, I got 2 cpus with only 2Gb of memory allocated. could this be too little memory?

[fpusan]

Yes, SqueezeMeta thinks you have 220.94 Gb available! :)
2 Gb is probably too little memory in any case, but you can control how much memory will be consumed by DIAMOND through the -b parameter.
See the documentation for tips & tricks when working in a computing cluster.