Comparing SV calls between methods

[peterdfields]

I would like to compare SV calls between two different methodologies, for example between a graph-based method and one which uses only a single linear reference. Is it reasonable to use sveval for this sort of purpose or is a 'truth-set' necessary (i.e., both the graph and linear reference methods are compared to the truth-set, not to each other)? Thank you for your time and advice.