Parameters on the left side of the model are displayed in their matrix form.
Variable notations in bivariate SEM-PGS model
Path coefficient matrices in bivariate SEM-PGS model
Variance and covariance matrices in bivariate SEM-PGS Model
Parameter Setup of Simulations
Median within-trait parameter estimates (+/- median absolute deviation, or MAD) as a function of Ntrios when = .04.
The red dashed line represents the true parameter values. Minor deviations between the median estimates and the true values were likely due to smal l deviations inrmate introduced during the bivariate foward-time simulation.
Parameter Setup of Simulations for Bias Analysis of Univariate Model
Median cross-trait parameter estimates and their MAD as a function of the Ntrios, when = .04.
MAD of within-trait parameter estimates as a function of Ntrio when = .04.
The red line represents the MAD when the latent genetic path coefficient a was freely estimated, while the blue line represents the MAD when a was fixed at its true population level.
MAD of within-trait parameter estimates are shown as a function of for trait 1, when N = 32k.
See the caption of Figure 4 for additional details.
MAD of cross-trait parameter estimates are shown as a function of number of trios (N), when the = .04.
Distribution of six key within-trait parameter estimates from a sensitivity analysis where the direct genetic effect, a11, was fixed to various values (Ntrio = 32k and = 4%).
Each violin plot shows the distribution of estimates across 100 replications fitted in OpenMx. The red dashed line highlights the results obtained when a11 was fixed to its true simulated value.
Distribution of six key cross-trait parameter estimates from a sensitivity analysis where the direct genetic effect, a11, was fixed to various values.
See the caption of Figure 9 for additional details.
Histograms of 500 parameter estimates from a univariate SEM-PGS model when the data was simulated with strong cross-trait effects.
The red dashed line indicates the true parameter value. The green dashed line indicates the median of the 500 estimates. P-values of the bootstrap test for each parameter are shown for all parameters. Omega is the covariance between parental phenotype and one haplotypic PGS. Gamma is the covariance between parental phenotype and one haplotypic LGS.
Histograms of within-trait parameter estimates obtained by fitting the model to 500 datasets simulated from a multivariate normal distribution.
The red dashed line indicates the true parameter value, while the blue dashed line represents the median of the 500 estimates. P-values from bootstrap tests assessing whether the median significantly deviates from the true value are shown in the top-right corner of each panel.
Histograms of cross-trait parameter estimates obtained by fitting the model to 500 datasets simulated from a multivariate normal distribution.
The red dashed line indicates the true parameter value, while the blue dashed line represents the median of the 500 estimates. P-values from bootstrap tests assessing whether the median significantly deviates from the true value are shown in the top-right corner of each panel.
Histograms of 500 parameter estimates from a univariate SEM-PGS model when the data is simulated with strong within-trait effects and weak cross-trait effects.
The red dashed line indicates the true parameter value. The green dashed line indicates the median of the 500 estimates. P values of the bootstrap test for each parameter are shown for all parameters.
Distribution of VF11 and VF12 from a sensitivity analysis where the direct genetic effect, a11, was fixed to various incorrect values.
Each violin plot shows the distribution of estimates across 100 replications fitted in OpenMx. The red dashed line highlights the results obtained when a11 was fixed to its true simulated value. The underlying data were generated with a sample size of Ntrio = 32k and = 4%.
