Skip to main content

Table 6 SNV results for HG003 WGS, Hybrid

From: HELLO: improved neural network architectures and methodologies for small variant calling

  DeepVariant HELLO
Precision Recall Errors Precision Recall Errors
20 × 15x 0.998807 0.998758 8104 0.998937 0.998999 6872
20 × 30x 0.998882 0.998790 7748 0.999233 0.999186 5266
20 × 60x 0.999035 0.998691 7565 0.999418 0.999234 4487
30 × 15x 0.999043 0.998848 7018 0.999061 0.999022 6382
30 × 30x 0.999081 0.998938 6594 0.999306 0.999185 5022
30 × 60x 0.999151 0.998820 6752 0.999443 0.999248 4358
40 × 15x 0.999185 0.998838 6579 0.999165 0.998970 6209
40 × 30x 0.999193 0.999006 5993 0.999354 0.999170 4912
40 × 60x 0.999223 0.998893 6269 0.999473 0.999242 4277
50 × 15x 0.999286 0.998796 6383 0.999243 0.998915 6131
50 × 30x 0.999272 0.999035 5635 0.999387 0.999130 4937
50 × 60x 0.999244 0.998932 6071 0.999500 0.999235 4211
  1. Values in bold indicate the best operating point for each tool for the corresponding metric