Figure 3

Soft-clipped reads vs. improperly aligned reads at a structural variant breakpoint. The reference sequence is indicated in blue. Green bases are alignment matches, pink bases are mismatches, yellow bases are skips, and gray bases are soft-clipped. All four reads are sampled from the variant sequence and span the breakpoint, but only the top two are properly soft-clipped. The bottom two reads are aligned with mismatches and skips because the portion that should be soft-clipped is only a few bases long and the global alignment is used instead. Inaccurate soft-clipping can lead to false negatives for split-read based SV prediction algorithms and also lowers the accuracy of SHEAR’s heterogeneity estimation algorithm.