GitHub
08/19/2026, 12:11 PM--orf_caller values in one execution).
The pipeline's multi-caller pipeline test needs --diamond_dbs on to catch a real class of bug (a Nextflow .join() silently dropping duplicate left-side keys when multiple ORF callers share the same diamond db name), but combining it with the full two-row samplesheet/diamond_dbs.csv pushed that CI job over the runner's disk budget (confirmed via repeated CI failures, "no space left on device").
diamond_dbs_single.csv is the same ncbi-refseq-test row already used in diamond_dbs.csv -- no new database files, just fewer rows -- as the only row, chosen since it also exercises the parse_with_taxdump code path. Still enough to reproduce the duplicate-key bug (two callers sharing one db name) at roughly half the diamond-side CI cost.
nf-core/test-datasetsGitHub
08/19/2026, 1:36 PMGitHub
08/20/2026, 12:21 PM