Fig. 7

Effect of sequencing error. Histograms of r-scores similar to those in Fig. 6. Reads of lengths 450−500 bases were corrupted at random before classification. In the upper panels 1% and 5% of the bases in each read were corrupted with a base different from the correct one. In the lower panels insertions and/or deletions of lengths 5 and 10 bases were distributed randomly at 1% of the positions in each read