Hello everyone. What is the best practice for tools that need a database to be downloaded? Should we create two modules like tool/database_download and tool/analysis?

As long as that "database" is able to be saved locally and passed to the tool like that, then this is a very common pattern in nf-core. See eg VEP, AnnotSV

What Arthur says - but the second module for downloading should ideally be from a specific subcommand of the tool (not a custom series of wget/curl commands etc.)

Yes! Thank you both