Table 1.
Comparison of variant calling results of DeepVariant and DeepVariant-on-Spark with different combinations of CPUs/GPUs.
Variant calling pipeline | Variant type | CPUa | GPUb | F1c | Recall | Precision | True positive | False negative | False positive | Genotype mismatch | Total number of SNV calls |
---|---|---|---|---|---|---|---|---|---|---|---|
DeepVariant | SNP | 16 | 0 | 0.99940 | 0.99937 | 0.99943 | 3040855 | 1928 | 1744 | 363 | 3886287 |
32 | 0 | 0.99940 | 0.99937 | 0.99943 | 3040856 | 1927 | 1744 | 363 | 3886337 | ||
64 | 0 | 0.99940 | 0.99937 | 0.99943 | 3040856 | 1927 | 1744 | 363 | 3886366 | ||
96 | 0 | 0.99940 | 0.99937 | 0.99943 | 3040855 | 1928 | 1744 | 363 | 3886339 | ||
16 | 1 | 0.99940 | 0.99937 | 0.99943 | 3040855 | 1928 | 1744 | 363 | 3886287 | ||
16 | 4 | 0.99940 | 0.99937 | 0.99943 | 3040855 | 1928 | 1744 | 363 | 3886287 | ||
32 | 2 | 0.99940 | 0.99937 | 0.99943 | 3040856 | 1927 | 1744 | 363 | 3886337 | ||
64 | 4 | 0.99940 | 0.99937 | 0.99943 | 3040856 | 1927 | 1744 | 363 | 3886366 | ||
DeepVariant-on-Spark | 32 | 0 | 0.99940 | 0.99937 | 0.99943 | 3040856 | 1927 | 1744 | 363 | 3886403 | |
64 | 0 | 0.99940 | 0.99937 | 0.99943 | 3040856 | 1927 | 1744 | 363 | 3886403 | ||
128 | 0 | 0.99940 | 0.99937 | 0.99943 | 3040856 | 1927 | 1744 | 363 | 3886403 | ||
32 | 2 | 0.99940 | 0.99937 | 0.99943 | 3040856 | 1927 | 1744 | 363 | 3886403 | ||
64 | 4 | 0.99940 | 0.99937 | 0.99943 | 3040856 | 1927 | 1744 | 363 | 3886404 | ||
128 | 8 | 0.99940 | 0.99937 | 0.99943 | 3040856 | 1927 | 1744 | 363 | 3886403 | ||
| |||||||||||
DeepVariant | Indel | 16 | 0 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868527 |
32 | 0 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868535 | ||
64 | 0 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868520 | ||
96 | 0 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868535 | ||
16 | 1 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868527 | ||
16 | 4 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868528 | ||
32 | 2 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868535 | ||
64 | 4 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868520 | ||
DeepVariant-on-Spark | 32 | 0 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868541 | |
64 | 0 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868541 | ||
128 | 0 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868541 | ||
32 | 2 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868542 | ||
64 | 4 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868542 | ||
128 | 8 | 0.96168 | 0.95711 | 0.96628 | 478265 | 21432 | 17373 | 11151 | 868541 |
aCPU means the number of CPU cores. bGPU means the number of NVIDIA Tesla P100 GPUs. cF1 means F1 score calculated by 2∗(recall∗precision)/(recall + precision).