Issue with understanding usage of DBSCAN

[holtgrewe]

Maybe the use case has not been implemented yet, but here is my problem description. Can this be resolved with rusty-machine?

I have a set of objects (2D intervals) and a similarity measure (their Jaccard distance). I would like to cluster the intervals based on this (e.g., specify a sparse distance matrix). Is this supported?

Would this be supported if the triangle inequality does not hold?

Thanks