A unified framework for the estimation of polygenic direct effect, indirect effect, and GxE interactions in case-parent trio studies
Usage
PGS.TRI(
pgs_offspring,
pgs_mother,
pgs_father,
GxE_int = FALSE,
parental_indirect = FALSE,
formula = ~envir1 + envir2 + factor(s1),
E,
side = 2,
smalltriosize = FALSE
)Arguments
- pgs_offspring
The PGS values of the affected probands (children). A vector of length N, no missing values are allowed
- pgs_mother
The PGS values of mothers that corresponds to the children. A vector of same length N, no missing values are allowed
- pgs_father
The PGS values of fathers that corresponds to the children. A vector of same length N, no missing values are allowed
- GxE_int
Whether there are interaction effect between pgs and environmental variables that are of interest in the model. If FALSE, then "formula" and "E" are ignored.
- parental_indirect
Whether to estimate potential parental indirent effect, returns an estimated difference of mother and father parental effect (delta_MF = beta_M - beta_F).
- formula
The environmental variables of interest for the PGSxE interaction effect
- E
The environmental variables of interest for interaction effect. A vector of length N for one environmental variable or a data frame/data matrix of NxP for P environmental variables are allowed.
- side
Sided of the Wald test or t test, default is 2-sided.
- smalltriosize
Whether number of trios is small (<100), if TRUE, a t test will be used rather than a wald test.
Value
A list of results of PGS.TRI
- Coefficients_direct
Results of direct PGS effect, if GxE_int is TRUE, then the result will also include PGSxE interaction effects
- Coefficients_indirect
Results of indirect parental PGS effect difference: PGS_mother - PGS_father
- var_fam
Within-family variances for each family
- var_fam_sum
Sum of within-family variances
- log_LC
Log likelihood of the offspring's transmission component
- log_LP_profile
The log profile likelihood of parents' component, note that we can report this likelihood when only considering direct effects in the model
- log_likelihood
Results of the final log-likelihood. This is calculated as the sum of the offspring's and parents' components
- vcov
Variance-covariance matrix of coefficients for direct genetic effects and gene–environment interaction terms (PGS×E)