The way that nf-test is implemented in the dev branch is not using the full functionality of the nf-test framework.
Only a subset of files is being checked for their md5sums and for existence.

In the future, the nf-test files should list all files that are generated by the pipeline and their software versions and only test for md5sums of stable files.

For the 4 profiles test_all, test_mouse, test_taxa_specific_assembly, test_taxa_only that download data from NCBI/Entrez, one should update the input files for the test profiles such that they use the testdata with assembly ids: i.e. https://github.com/nf-core/test-datasets/blob/metapep/testdata/taxids_assemblyids.cond1.tsv instead of https://github.com/nf-core/test-datasets/blob/metapep/testdata/taxids.cond1.tsv.