Table 5 Details of the hyperparameters used in the normal and compositional simulations.
From: Edge and modular significance assessment in individual-specific networks
Normal distribution | Compositional | ||||
|---|---|---|---|---|---|
Parameters | Values | Details | Parameters | Values | Details |
N | 100, 500, 1000, 2000 | Controls + cases observations | N | 100, 500, 1000 | Controls + cases observations |
M | 1, 5, 10 | Cases observations | M | 1, 5, 10 | Cases observations |
k | 2, 3, 5, 7, 9, 11, 17 | Module’s size | k | 2, 5, 11, 17 | Module’s size |
Outlier generation | Common, Specific | Common: all outliers share a common distribution Specific: each outlier has a different variance-covariance structure. | Data heterogeneity | Uniform, \(\alpha\) = 4, \(\alpha\) = 0.7 | Degree of heterogeneity of the parameter to generate the data, going from no heterogeneity (Uniform) to high heterogeneity (Pareto with \(\alpha\) = 0.7 ) passing through mild heterogeneity (Pareto with \(\alpha\) = 4 ) |
Mult | 1.1, 1.5, 2 | Multiplying factor applied to a percentage of observation to differentiate between cases and controls observations | |||
Percentage increase | 10%, 25%, 40% | Percentage of inflated parameters on the total differentiating cases and controls | |||
