Hey @Lu_Yang,
We actually have a bunch of features for joined data which will be in the next release. But basically, yes that will work soon, you'll just be specifying a slightly different semantic type (SampleData[JoinedSequencesWithQuality]).
DADA2 works best if it can denoise the forward and reverse reads independently before merging. It can kind of work with joined data, but it requires the overlapping quality scores to basically match the profile of the quality scores around it and that really depends on what read-joiner you used. (Then there's also the typical trimming and removal of non-biological sequence requirements.)
I would recommend waiting for the 2017.11 release where there should be a few other analysis options that you can use for this data.