Filtering a result based on some deeper feature in the graph is a really powerful thing. Problem is the level of nesting needed is unbounded, so that ends up making a really big schema and really deep queries.
Adding onto what @michaelcomptonsaid, you can also check out the section on cascade in this blog https://blog.dgraph.io/post/tutorial-8-getting-started/ . Cascade in Dgraph’s GraphQL± query language helps you filter the top layer results based on the filter in the deeper parts of the query.