I only have one database, containing my main tree (my children, all their ancestors and all relatives of ancestors) and a number of unconnected trees.
I can't see many advantages to having separate files for different lines. My software allows for reporting and export of any given subset of my tree...
...and if you had, say, two separate databases for different lines, would you not have to maintain all repositories, sources, citations, etc., twice over? Sounds like a recipe for inconsistency and duplication of effort. And what if one individual happens to appear in both lines?