I was able to gain access to a different cluster and it's going further, but now it's erroring out in the SPAdes assembly. Is there something I can change that can fix this?

Glad that you didn't give up on the pipeline! 🙂 You're using Spades which requires much more memory than MEGAHIT, so my first recommendation is to switch to MEGAHIT (

--assembler megahit

). Second, use the latest release of the pipeline (1.5):

nextflow pull nf-core/megahit -r 1.5
nextflow run nf-core/megahit -r 1.5 --assembler megahit <other params>

There's a considerable risk that the default memory allocation for MEGAHIT won't suffice either. Ping us again if that happens.

Thank you! Unfortunately, I am currently unable to update the version (it's a long process). However, your comment regarding memory reminded me that running SPAdes has always required over 500 gb ram. I was only allowing 128 gb. So I have fixed that and it seems to be running okay.

Aren't you using containers, i.e.

-profile docker

or

-profile singularity

(

-profile conda

also possible)? Through that you don't have to have Megahit, it's pulled in for you.

I'm not sure? I'm running it through my institution’s system which is a bit of a closed door. They looked into the program and loaded it up, and the submission system is more GUI rather than command line, which is strange but it’s mostly working. I'm helping debug; so I'll bring that up if SPAdes (which I do prefer in any case) can't run

Using the

-profile

param to Nextflow is quite standard. And for most larger computers, Singularity or Apptainer is what's used, so ask the admins about that. Using that means you don't have to install any software for the actual compute steps. Also, regarding your earlier comment on which version is available: The normal procedure is to let Nextflow pull down the pipeline code. So, it's simple as long as you have Nextflow and Apptainer, just

nextflow run nf-core/metatdenovo -profile singularity <your pipeline params>

and it will run. In case the machine you're running this on doesn't have internet access there's a prepare for run command in the nf-core tool. If that's relevant, ping me again. How this plays out with the GUI you mentioned, I have no idea, but this might be worth passing on to admins.