Jonathan Manning

Ahh, --assets is what I was missing- thanks both. I can make sure the genome fasta gets built first.

For anyone finding this and wondering how to attach cDNA FASTAs to a genome my query above might mislead you. To correct the record, you should actually do e.g. :...

Thanks for the response- I wondered about the DB backing too so glad to see there's an issue for that.

Having done some more testing, I wanted to re-iterate that this is actually pretty problematic at scale, at least in our compute environment. For example I've noticed that occasionally, while...

Thanks @nsheff . I'm parallelising using a Nextflow workflow (see https://github.com/ebi-gene-expression-group/isl_refs_to_refgenie), pointing at our LSF cluster.

@nsheff for the dependencies: yes of course, that's pretty much the whole point of composing this as a workflow :-). The workflow structure encodes that, so e.g. the outputs of...

Just a note: if I were writing the above workflow again I'd probably use Snakemake which would allow some of that fiddly Nexflow logic to be removed.

Just an addition to this issue: when I retry an asset build job after random failures, I add the '-R' flag (I'd found that locks from failed builds sometimes prevent...

If it helps, here's some more illustration. The issue impacts more than the species wit the problem. For example my workflow was trying to build the base carrot genome assembly...

@nsheff I'm building and indexing the cDNAs as instructed in https://github.com/refgenie/refgenie/issues/250, so both under 'fasta', but it does seem to work on a small scale, as per my example at...