Targeted Validation of Somatic Variants

To validate variants identified by MuSE and Caller A in the ACC data, we selected 550 patient-specific positions and designed NimbleGen probes correspondingly for the purpose of targeted capture enrichment and deep sequencing. Paired-end Illumina resequencing was carried out to an average sequencing depth at 1500×. After mapping the reads against the hg19 reference genome using BWA, we considered a somatic variant as validated if its p value calculated from Fisher’s exact test comparing the tumor and matched-normal samples was not larger than 0.05. The validation rates of MuSE and Caller A were calculated as
validation rate of MuSE unique calls
$\begin{array}{c}=\frac{1}{139+2303}\left[\frac{8}{11}·(11+141)+\frac{7}{39}·(39+221)\right.\\ \left.+\frac{5}{34}·(34+345)+\frac{8}{25}·(25+494)+\frac{7}{30}·(30+1102)\right]\\ \approx 0.2634,\end{array}$
validation rate of MuSE shared calls
$\begin{array}{c}=\frac{1}{290+9584}\left[\frac{125}{125}·(125+8900)+\frac{99}{111}·(111+472)\right.\\ \left.+\frac{25}{29}·(29+109)+\frac{12}{17}·(17+52)+\frac{7}{8}·(8+51)\right]\\ \approx 0.9889,\end{array}$
validation rate of MuSE total calls
$\begin{array}{c}=\frac{1}{139+290+2303+9584}\\ \times \left[0.2634·(139+2303)+0.9889·(290+9584)\right]\\ \approx 0.8450,\end{array}$ $\begin{array}{c}\text{validation rate of Caller A unique calls}\\ =\frac{30}{121}\\ \approx 0.2479,\end{array}$ $\begin{array}{c}\text{validation rate of Caller A total calls}\\ =\frac{1}{121+290+1693+9584}\\ \times \left[0.2479·(121+1693)+0.9889·(290+9584)\right]\\ \approx 0.8739.\end{array}$