I was expecting that, when running minimap2 against the same input files, the alignments output into paf (ie without the -a option) would be the same as those output into sam (ie using the -a option). However, this appears not to be the case. Is this expected behavior? Why does it seem like I'm getting longer end-to-end alignments in the paf output compared to the sam output (where the alignments end up getting split with multiple supplementary alignments)? What happens to supplementary alignments when outputting to paf?

(I'm using 2.10-r764-dirty to map very long ont DNA reads.)