We were wondering how LRA handles ambiguity in the alignment as there are no secondary alignment outputs.