Short reads cannot distinguish two repeated copies, but paired reads show that one read maps to unique sequence on one side and its partner maps beyond the repeat. How can this information help?