Skip to main content
. 2017 Nov 21;7:15959. doi: 10.1038/s41598-017-14896-7

Table 3.

The averaged performances of GATK and GATKcan in detecting mutations from (A) ~61,507 variants of randomly sampled 231 endometrial tumors in ten repeats, and (B) ~52,291 variants of randomly sampled 197 endometrial tumors in ten repeats, checked against the 184,824 reported mutations in EC of TCGA.

GATK GATKcan
A.
Training TPR§ (s.e.) -----*----- 99.0 (1.0)
cFPR (s.e.) -----*----- 11.9 (1.3)
Test TPR (s.e.) 88.2 (0.6) -----*----- 96.1 (1.9)
cFPR (s.e.) 65.2 (0.1) -----*----- 12.2 (1.6)
precision (s.e.) 12.6 (0.4) -----*----- 46.0 (3.2)
MCC (s.e.) 14.5 (0.5) -----*----- 61.8 (2.0)
B.
Training TPR§ (s.e.) -----*----- 98.8 (0.5)
cFPR (s.e.) -----*----- 12.1 (1.6)
Test TPR (s.e.) 88.1 (1.3) -----*----- 96.1 (2.1)
cFPR (s.e.) 64.9 (0.4) -----*----- 12.3 (1.6)
precision (s.e.) 12.6 (0.9) -----*----- 45.6 (2.9)
MCC (s.e.) 14.6 (1.0) -----*----- 61.4 (1.8)

§The unit of all performance measures and their s.e.’s are %. *Denotes the P value of the two sample t-test < 10−7.