How to use paired-end library information to confirm an organelle genome configuration with repeats larger than read length/largest k-mer?

[Kinggerm]

[Accipiterbremanii]

Would it be possible to subselect large scaffolds first or set a scoring value for deciding when deciding which scaffold should be connected to which? Or introduce some pre existing knowledge on the general layout of the plastome. Usually they are LSC-IR-SSC-IR (long single copy, inverted repeat, short single copy, inverted repeat). Which could be used to make a more informed 'guess' as to which scaffold should go where?. What I do now with the output is to identify these main regions based on temporary assembly results. These seem to be ok. I terminate the run before the final connections are made.. And loose some information of course.