Hello all. I'm at my wits end trying to debug running metadenovo, as it keeps erroring out and I cannot find where the error is coming from. I have attached my .log file and my nohup file from running the command: nohup nextflow run nf-core/metatdenovo -profile conda -r 1.3.0 -c local_config.config --input samplesheet.csv --assembler megahit --orf_caller transdecoder --eukulele_dbpath /home/ec2-user/s3-drive/uploaded-files/eukulele --eukulele_db phylodb --outdir ~/s3-drive/output & I put an additional config file in with:

// local_config.config
process {
    resourceLimits = [ cpus: 32, memory: 240.GB, time: 72.h ] // Set to your server specs
}

I downloaded the phylodb databases separately and have them accessible. I am working through AWS and am using a 32 core 256 GB instance.

Is there any particular reason why you're running with

nohup

? Using

screen

you can run the command in the foreground, detach from the session, logout and it will continue running until you login and reattach to the session. (This also means that you shouldn't add

&

after the command.)

I'm using AWS console and entering instances to run, so I'm not sure if its using Uppmax or not

Right, probably not. I'm unfamiliar with AWS. In any case, do you think you can run with either Singularity or Docker? Conda tends to be brittle.

I've tried to get Docker set up, but I've had a lot of difficulty and no success in getting it to work (I went with conda because it installed when docker didn't)

And Singularity/Apptainer? I'm trying to understand how you run this. From your command above it looks like just local execution except for the output directory. Is "AWS console" just a virtual machine in which you install whatever software you need and then run like in a local machine? (Sorry for my inexperience with AWS.)

I haven't tired singularity, I could try that. The AWS console uses whats called an instance where you enter a machine with a specific number of cores and RAM. I'm running it locally on that machine with the available cores and RAM, rather than submitting it to a cluster

I see, that's what I thought. Would be great if you can try Singuliaryt/Apptainer and then run Nextflow in the foreground in a Screen session so we hopefully get better error messages.

okay, I can see if that can work. I'll try to run Nextflow first in the foreground to see what I get

Yes:

-w path

thank you!