S-turns (a.k.a. bulged-G motifs, loop E motifs, sarcin/ricin loop)

[rhiju]

There are noncanonical motifs that recur inside internal loops (e.g., in ribosome sarcin/ricin loops, glutamine riboswitch RNA puzzles, etc.)

...GUAX-3'
    AGY-5'

where X/Y is a Watson-Crick pair

See, e.g.,

This kind of motif needs to be handled differently than, say, a K-turn, which is a complete internal loop with flanking base pairs at both ends.

[rhiju]

See also #24 -- the implementation of this feature would be unified with the dangles/mismatches.