Strand-asymmetry of high frequency indels was only evaluated on the indel polymorphisms found in the 200 bp kits (both Manual and OneTouch), and only for sites that had reads mapping in both orientations. For each site, the data form a 2×2 table, namely strand by presence/absence of indel. We therefore used Fisher's exact test to generate a p-value for every site, testing the null hypothesis that indel frequency at that site was not orientation specific. The p-values were then adjusted to control the family-wise error rate using Holm's method [22]. Only sites with corrected p-values smaller than 5% were considered to demonstrate significant evidence of orientation specific indel frequencies at that site. Using this approach, the orientation specificity may differ by site. (The analysis was repeated using a chi-squared test in place of Fisher's exact test with similar results).
