mitochondrial splice junctions detected by STAR

[asilvestris84]

Dear Alex, I've decided to write here, first of all, for clarification on why STAR, when launched in two-pass basic mode, identifies splice junctions on the mitochondrial genome in the first scan (which to the best of my knowledge at least in human makes no sense from a biological perspective). Secondly, I want to know if it would be possible to insert a filtering step between the first and second passes that discards these obviously incorrect junctions. I look forward to your response, even just to the first question (perhaps the mitochondrial reads should be mapped separately with a non-splice-aware aligner?). Thanks in advance. Ale