On 02/28/2014 09:47 AM, Hervé Pagès wrote: > Hi Chris, > > On 02/27/2014 11:00 PM, chris warth wrote: [...] >> Why in the world does coverage() care if the alignments have strand >> metadata? > > It's not coverage() that cares. It's the internal coercion to > GRangesList that happens internally before the coverage is actually > computed. This internal coercion tries to propagate the metadata > columns, but, for reasons I never really understood, GRanges and > GRangesList objects are not allowed to hold a metadata column named > "strand". The fix is easy: this internal coercion to GRangesList should > not propagate the metadata columns (they're not needed for computing the > coverage and they slow down the coercion). I'll fix this today. This is fixed in GenomicAlignments 0.99.27. Cheers, H.