Table 2.
Root-mean-square error measurements for methods on simulated data
| Dataset type | Base dataset | k | n | m | ADMIXTURE | fastStructure | TeraStructure | ALStructure | sNMF | SCOPE |
|---|---|---|---|---|---|---|---|---|---|---|
| PSD | HGDP | 6 | 10,000 | 10,000 | 4.0∗ | 10.3 | 16.6 | 5.6 | 4.1 | 5.6 |
| PSD | TGP | 6 | 10,000 | 10,000 | 1.8∗ | 15.9 | 13.7 | 3.2 | 4.1 | 3.2 |
| PSD | TGP | 6 | 10,000 | 1,000,000 | 0.2∗ | 12.4 | 0.9 | – | – | 0.3 |
| PSD | TGP | 6 | 100,000 | 1,000,000 | – | – | 1.0 | – | – | 0.4∗ |
| PSD | TGP | 6 | 1,000,000 | 1,000,000 | – | – | – | – | – | 0.5∗ |
| Spatial | HGDP | 6 | 10,000 | 10,000 | 11.9 | 31.1 | 10.2 | 5.7∗ | 5.7∗ | 6.5 |
| Spatial | TGP | 6 | 10,000 | 10,000 | 12.5 | 29.1 | 6.8∗ | 7.5 | 9.4 | 7.3 |
| Spatial | TGP | 10 | 10,000 | 100,000 | 10.8 | 22.8 | 8.8 | 8.5 | 6.7∗ | 6.7∗ |
| Spatial | TGP | 10 | 10,000 | 1,000,000 | – | – | 6.6∗ | – | – | 7.2 |
Root-mean-square error (RMSE) was computed against the ground truth admixture proportions for each simulation. RMSE is displayed in percentage and rounded to the first decimal place. A dash denotes that the method was not run due to projected time or memory usage. Values with an asterisk denote the best value for each dataset.