Trait-specific covariates

Formula interface in solarPolygenic is used differently than in standard functions like lm. The variable names are extracted from left-hand and right-hand sides of the formula argument, checked for existence in data columns and then passed to SOLAR. Thus, formula is such that SOLAR likes it. That allows to specify covariates in a trait-specific manner like sex(trait1).

Common covariates

mod1 model is created with a common formula format, that means sex covariate is given for both traits trait1 and trait2.

mod1 <- solarPolygenic(trait1 + trait2 ~ sex, dat30)

Trait-specific covariates

Imagine that sex is the only significant covariate for trait1, and age is the only significant covariate for trait2.

trait covariates
trait1 sex
tarit2 age

Covariates in mod2 model are trait-specific.

mod2 <- solarPolygenic(trait1 + trait2 ~ sex(trait1) + age(trait2), dat30)

Mix of common and trait-specific covariates

A mix of common and trait-specific covariates is also possible.

trait common covariates specific covariates
trait1 sex  
tarit2 sex age 
mod3 <- solarPolygenic(trait1 + trait2 ~ sex + age(trait2), dat30)

traits and covlist arguments

args(solarPolygenic)

## function (formula, data, dir, kinship, traits, covlist = "1", 
##     covtest = FALSE, screen = FALSE, household = FALSE, alpha = 0.05, 
##     polygenic.settings = "", polygenic.options = "", verbose = 0, 
##     ...) 
## NULL

mod2.par <- solarPolygenic(traits = c("trait1", "trait2"), 
  covlist = c("sex(trait1)", "age(trait2)"), data = dat30)

Covariates for mod2.par model are trait-specific.