Table 1 SomaticSeq sensitivity on real data. ≥N tools represents the consensus of at least N callers. The two sets of real tumor–normal sequencing data were downloaded from the European Genome Archive. SomaticSeq predictions were trained from the DREAM Stage 3 data sets
From: An ensemble approach to accurately detect somatic mutations using SomaticSeq
Sample | COLO-829 | CLL1 | ||
|---|---|---|---|---|
SNV | Number of calls | Sensitivity | Number of calls | Sensitivity |
MuTect | 46,831 | 0.996 | 8,361 | 0.895 |
VarScan2 | 64,927 | 0.987 | 19,797 | 0.888 |
SomaticSniper | 53,077 | 0.996 | 13,690 | 0.907 |
JointSNVMix2 | 85,983 | 0.996 | 22,534 | 0.899 |
VarDict | 53,076 | 0.857 | 5,748 | 0.883 |
Union of five tools | 191,696 | 0.998 | 55,196 | 0.935 |
≥2 tools | 57,254 | 0.996 | 13,594 | 0.916 |
≥3 tools | 43,848 | 0.996 | 5,836 | 0.904 |
≥4 tools | 38,216 | 0.996 | 2,637 | 0.886 |
=5 tools | 34,086 | 0.857 | 1,749 | 0.842 |
SomaticSeq | 37,452 | 0.996 | 2,320 | 0.892 |