Illegal instruction error singlem_pipe_reads

[Anna-MarieSeelen]

Hi Binchicken developers,

I'm testing out binchicken on our cluster, but I keep getting an illegal instruction error from the first rule, called singlem_pipe_reads. This is the command I'm trying to run:

binchicken coassemble
--forward $RUNDIR/input/_1.paired.trim.fastq.gz
--reverse $RUNDIR/input/2.paired.trim.fastq.gz
--output $RUNDIR/output${SLURM_JOB_NAME}_${SLURM_JOB_ID}
--taxa-of-interest "g__Nitrobacter" --cores 10 --tmp-dir /scratch

And the illegal instruction error that I get is in the general log file:
example_bin_chicken_6311149_stderr.txt

The individual log files that I get from the singlem pipe rule look free from errors to me, one example below:
NB8_09-11_read.log

I'm unsure why its giving the illegal instruction error, and what I should do to solve it. Do you have any ideas on this?

Thanks in advance!

Best,

Anna

[wwood]

Hi @Anna-MarieSeelen thanks for the bug report.

I'll leave it to @AroneyS to give a fuller answer, but I note that in your log file it says

06/24/2025 07:28:04 PM INFO: SingleM v0.13.2

But then it goes on to act like a much newer version of SingleM, which is quite strange. I wonder if there is something strange in your environment somehow.

[AroneyS]

The SingleM version is odd. Bin Chicken v0.12.11 should be using SingleM v0.19.0.

What happens when you run SingleM manually, by activating ../../../vol/micro-bioinfo/Anaconda3-24/envs/2ca2f28a16cd94fdaf4e6e96f2cac6ca_ and running

singlem pipe \
--forward /scratch/example_bin_chicken-ej8/input/NB8_09-11_1.paired.trim.fastq.gz \
--reverse /scratch/example_bin_chicken-ej8/input/NB8_09-11_2.paired.trim.fastq.gz \
--otu-table /scratch/example_bin_chicken-ej8/output_example_bin_chicken_6311149/coassemble/pipe/NB8_09-11_read.otu_table.tsv \
--metapackage /vol/micro-databases/singlem/S4.3.0.GTDB_r220.metapackage_20240523.smpkg.zb \
&> /scratch/example_bin_chicken-ej8/output_example_bin_chicken_6311149/coassemble/logs/pipe/NB8_09-11_read.log

You will also run into a problem with sample names in later steps, since SingleM will use NB8_09-11_1.paired.trim (everything but fastq.gz from the 1st pair) but Bin Chicken will use NB8_09-11 (the common string between paired reads). Its currently fixed by removing _1 from the names that SingleM uses, but that wont work here since you have .paired.trim after the _1. Current fix is to rename inputs to the form [sample ID]_1.fastq.gz and [sample ID]_2.fastq.gz.

I'm aiming to fix this properly in a later version (by simply using SingleM's method instead).