Files I think we could (at least have a switch to) remove from publish:

COMEBin/stats/*-COMEBin-*/embeddings.tsv

40-70mb each, piles up quickly! about 7.3Gb for a 60x4 samples co-assembly; 29 Gb for 240 single assemblies;

Assembly/*/QC/*/ALE/*output.txt

takes 10-15 Gb per sample, does that data really need to be published? Seems preposterous to publish this by default, but is there a reason I'm not aware of?

Thanks for the COMEbin suggestion! And about the ALE, that's the thing I said earlier. ALE by default generates a file with general statistics + a row per position, and for metagenomes that's huge. In dev (and next version) this extended output is disabled by default using

--nout

, and can be turned on with a new parameter.

ah sorry, my short-term memory really sucks haha