Table 1 Performance comparison of several global ancestry inference algorithms
Dataset | Algorithm | Δ(Q, QGT) | RMSE(Q, QGT) | RMSE(F, FGT) | Runtime (CPU) | Runtime (GPU) |
|---|---|---|---|---|---|---|
All-Chms | ADMIXTURE | 0.042 | 0.153 | 0.062 | >1 day | – |
AlStructure | 0.064 | 0.159 | 0.032 | 06:04:28 | – | |
TeraStructure | 0.033 | 0.133 | – | 02:12:46 | – | |
HaploNet | 0.026 | 0.114 | – | – | 03:17:00 | |
Neural ADMIXTURE | 0.025 | 0.108 | 0.011 | 00:11:21 | 00:01:32 | |
Chm-22 | ADMIXTURE | 0.048 | 0.161 | 0.068 | 02:56:29 | – |
fastSTRUCTURE | 0.055 | 0.162 | – | 03:31:00 | – | |
AlStructure | 0.116 | 0.256 | 0.068 | 00:46:49 | – | |
TeraStructure | 0.050 | 0.170 | – | 00:43:48 | – | |
HaploNet | 0.053 | 0.170 | – | – | 01:09:29 | |
Neural ADMIXTURE | 0.033 | 0.140 | 0.016 | 00:05:46 | 00:00:45 | |
Chm-22-Sim | ADMIXTURE | 0.046 | 0.197 | 0.067 | 09:48:18 | – |
fastSTRUCTURE | 0.069 | 0.237 | – | >1 day | – | |
AlStructure | 0.126 | 0.286 | 0.076 | 02:51:36 | – | |
TeraStructure | 0.040 | 0.175 | – | 06:37:14 | – | |
HaploNet | 0.026 | 0.113 | – | – | 02:07:54 | |
Neural ADMIXTURE | 0.011 | 0.070 | 6.02 × 10−3 | 00:20:41 | 00:01:34 | |
PAB | ADMIXTURE | 1.44 × 10−4 | 0.010 | 5.97 × 10−3 | 03:31:01 | – |
AlStructure | 1.45 × 10−3 | 0.026 | 7.83 × 10−3 | 05:10:42 | – | |
TeraStructure | 1.97 × 10−4 | 0.012 | – | 01:13:38 | – | |
HaploNet | 0.039 | 0.248 | – | – | 02:37:09 | |
Neural ADMIXTURE | 4.34 × 10−3 | 0.055 | 7.01 × 10−3 | 00:14:27 | 00:01:48 | |
Synthetic | ADMIXTURE | 1.37 × 10−4 | 0.011 | 0.028 | 00:08:06 | – |
AlStructure | 2.74 × 10−4 | 0.014 | 0.030 | 00:03:07 | – | |
TeraStructure | 1.13 × 10−3 | 0.032 | – | 00:03:28 | – | |
HaploNet | 0.022 | 0.123 | – | – | 00:04:04 | |
Neural ADMIXTURE | 8.60 × 10−4 | 0.030 | 0.028 | 00:01:25 | 00:00:12 |
