This article claims it can detect differences at the single-nucleotide-difference level.
How does one make sure that a "difference" wasn't due to a mistranscription error during the typing? Or is the error rate from that so low that it is brushed aside?
Cheers!