A tandem simulation framework for predicting mapping quality

Abstract

Read alignment is the first step in most sequencing data analyses. Because a read’s point of origin can be ambiguous, aligners report a mapping quality, which is the probability that the reported alignment is incorrect. Despite its importance, there is no established and general method for calculating mapping quality. I describe a framework for predicting mapping qualities that works by simulating a set of tandem reads. These are like the input reads in important ways, but the true point of origin is known. I implement this method in an accurate and low-overhead tool called Qtip, which is compatible with popular aligners.