It probably makes sense to make these separate commands, since that way we can more easily tell which command fails (if that happens).
Also, we typically use sed -i rather than perl -i -pe:

postinstallcmds = [
    "rm -rf %(installdir)s/docker/ %(installdir)s/singularity/",
    "chmod a+rx %(installdir)s/*.pl",
    "chmod -x %(installdir)s/*.jar",
    "sed -i 's|helpfile = \"help\"|helpfile = \"%(installdir)s/help\"|' %(installdir)s/inparanoid.pl",
    ...,
]

can also use fix_perl_shebang_for = ['*.pl'] instead of the perl -i -pe 's|#!.*perl|#!/usr/bin/env perl|' %(installdir)s/*.pl command

perl is a requirement, that's why I used it instead of sed. I can change for sed if you want.

How to do separate accounts in the postinstallcmds ?

postinstallcmds is a list of strings; see boegel's example above

How about running this in the build directory instead, then cleanup is not needed:

sanity_check_commands = [
    "cd %(builddir)s && inparanoid.pl -input-dir %(installdir)s/testInput/",
]

Make sense

Also check for testInput here?

Right

@smoretti We should rename this on download:

version = '5.0.79d8e39'
local_commit = '79d8e39bb243d403ce57b699c1ae104a8a640389'
...
sources = [{'download_filename': '%s.tar.gz' % local_commit, 'filename': SOURCE_TAR_GZ}]

@smoretti I only notice now that this is a self-invented version...

I think it would be better to ensure we use a version that sorts well, so use a datestamp of the commit instead?

version = '5.0-20220607`