Limits in accuracy and a strategy of RNA structure prediction using experimental information