Figure 1: Principle of the indel detection algorithm. Sequence reads are processed using a rapid alignment program (e.g.: MAQ, or ELAND) configured for identifying only mismatches. Green lines represent sequence reads which align through an edge effect. Red lines represent reads which fail to align. The position of the indel is indicated by an arrow (insertion) or a box (deletion). A portion of the sequences align through their edges and are reported erroneously as containing multiple mismatches. Most of the reads with an indel located in the center of the sequence fail to align. This causes the depth to drop in the region flanking the indel and the mismatch rate to rise at its boundaries, as shown in the example plots. These plots are example regions taken from the phix174 spike-in study.