Table 2 The effect of parameter values on outlier detection using simulated data
From: Robust principal component analysis for accurate outlier sample detection in RNA-Seq data
Outlier added | Outlier model | Error rate | Method | Parameter name | Default (range) | Parameter value | Outlier called | Number FP outlier called |
|---|---|---|---|---|---|---|---|---|
L-1 | outlierL | 0.01 | PcaGrid | crit.pca.distances | 0.975 | ≤ 0.844 | Yes | >  1 |
0.845–0.912 | Yes | 1 | ||||||
0.913–0.999 | Yes | 0 | ||||||
1 | No | 0 | ||||||
PcaHubert | crit.pca.distances | 0.975 | ≤ 0.643 | Yes | >  1 | |||
0.643–0.982 | Yes | 1 | ||||||
0.983–0.999 | Yes | 0 | ||||||
1 | No | 0 | ||||||
alpha | 0.75 (0.5–1) | 0.5–0.683 | Yes | 1 | ||||
0.684–0.749 | Yes | >  1 | ||||||
0.750–0.999 | Yes | 1 | ||||||
1 | No | 0 | ||||||
H-1 | outlierH | 0.01 | PcaGrid | crit.pca.distances | 0.975 | ≤ 0.648 | Yes | >  1 |
0.649–0.750 | Yes | 1 | ||||||
0.751–0.999 | Yes | 0 | ||||||
1 | No | 0 | ||||||
PcaHubert | crit.pca.distances | 0.975 | ≤ 0.669 | Yes | >  1 | |||
0.669–0.978 | Yes | 1 | ||||||
0.979–0.999 | Yes | 0 | ||||||
1 | No | 0 | ||||||
alpha | 0.75 (0.5–1) | 0.5–0.714 | Yes | 0 | ||||
0.715–0.999 | Yes | 1 | ||||||
1 | No | 0 |