Table 1. Computational costs for the solution of single-trait and multiple-trait analyses.
The variables n, m and t denote the sample size, the number of genetic markers, and the number of traits, respectively. v is the average number of iterations necessary to estimate the model parameters σ 2 and h 2 (see “Time complexity” in Supplementary Note S1).
| Algorithm | Estimation
of σ 2 and h 2 |
Single-trait
analysis |
Multi-trait
analysis ( t > n) |
Multi-trait
analysis ( n > t) |
|---|---|---|---|---|
| CLAK-C hol | O( n 3 + tvn) | O( mpn 2) | O( tmn 2) | O( tmn 2) |
| CLAK-E ig | O( n 3 + tvn) | O( mpn 2) | O( tmn) | O( mn 2) |
| FaST-LMM | O( n 3 + tvn) | O( mpn 2) | O( tmn 2) | O( tmn 2) |
| GWFGLS | O( n 3 + tvn) | O( mpn 2) | O( tmn 2) | O( tmn 2) |
| EMMAX | O( n 3 + tvn) | O( mpn 2) | O( tmn 2) | O( tmn 2) |