To eliminate these additional false positives due to inaccurate read placement and non-independent sequencing errors, we developed six filters (Fig. 1, Table 1). In addition, we use a panel of normal samples as controls to further eliminate both germline events and artifacts (Online Methods). Subsets of these filters define several versions of the method (Fig. 1): (i) Standard (STD), which applies no filters and thus includes all detected variants; (ii) High Confidence (HC), which applies the six filters and (iii) High Confidence + Panel of Normals (HC+PON), which additionally applies the Panel of Normals filter.
