If there is one type of node that is common between two databases, should they always be one database?

I am looking at the possibility of creating a graph database or multiple databases using neo4j. There are two or more very distinct purposes for these databases, and they are not related to each other except that a type of node common between them. Would it be better to use separate databases that just have a type of node in common with each other, or a single database where the node in question has multiple types of relations/edges connecting to nodes that otherwise have no relation nor any purpose in relating to each other? Also, each user will only have a single use case applicable to them, so the other use cases would not be relevant to them (the extra relation types and nodes will have no purpose for them). I am looking at this from multiple perspectives, including performance, storage space, memory usage, ease of creation, and maintenance. I am inclined to view a single database as being the best solution, partly based on past experience (I created a neo4j database for NASA that is part of one of their directorate's production environment now), but I could be wrong.

I would create two separate databases specific to the individual requirements. This would eliminate dependencies between the projects, making development, maintenance, and availability easier to address for each. Also, there may be overlap in the data models today, but they may diverge as additional requirements and learnings are gained.