Hello, This is only my opinion right...
But it seems to me that converting hbonds into secondary structure elements (helices, strands, paperclips...) can be tricky at the boundaries when bonds are just too short or just the other side of the correct angle. To illustrate this with 1m6q (which is a model, by the way, and has some odd features, like not-quite-bonded sheets) I pick one of the short sheets you circle (at least I think I do) : 379-383::423-427.
