Figure 8.

The paired t-test statistic helps distinguish true SNPs from systematic errors. The paired t-test (PT(w0, w1)) was computed for the "SNPs" and "Systematic errors" sets used for training SysCall. The histogram of paired t-test for the "SNPs" set (red) is centered around 0 (mean: 0.0024, std: 2.035), indicating that the quality scores at those locations were similar to their neighboring quality scores. The histogram of the "Systematic errors" set (blue) formed an almost disjoint distribution (mean: -10.505, std: 3.919).

Meacham et al. BMC Bioinformatics 2011 12:451   doi:10.1186/1471-2105-12-451
Download authors' original image