createSimModel: Create strumSimModel Object

Description

A function to create an object of class strumSimModel which contains the information used for simulation.

Usage

createSimModel(formulas, tMissingRate = c(0), markerInfo = NULL, ascertainment = NULL, defaultError = '', assumeExogCovariate = TRUE)

Arguments

formulas

Vector of expressions that defines the relationship among the variables in the model with the fixed coefficient (see Details).

markerInfo

Object containing a strumMarker or NULL.

tMissingRate

Numeric vector to hold the missing rate(s) of the simulated trait(s) in the model. The length of vector is equal to the number of observed traits or 1 if one missing rate is applied to all traits.

ascertainment

Function stating the ascertainment criteria of the data (see Details).

defaultError

Vector of chractor listing the default variance components in the model (see Details).

assumeExogCovariate

Logical. If TRUE, then it is assumed that any observed, quantitative exogenous traits are covariates. Otherwise, they are treated as latent and dependent variables.

Value

Returns an object of type strumSimModel.

Details

This function is used to create a strumSimModel object which will be an input parameter for simulateStrumData function.

The formulas argument is given as a charactor string of several expressions that defines the relationship among the variables. Blank lines and comments (line start with #) can be included between formulas.

Note that the coefficients of the right hand side variables of the measurement equations are fixed with the value given in the formulas.

Three different equations are allowed in the formulas syntax:

Measurement equations

The "=~" operator may be read as "measured by" and it specifies the measurement equations in the model. The left hand side of the "=~" must contain one unobserved or latent factor variable name. The right hand side of the "=~" are observed variables, observed covariates, and/or any measurement errors enclosed with "<>".

Structural equations

The "~" operator specifies the structural equations in the model. The left hand side of the "~" must contain one unobserved or latent factor variable name. The right hand side of the "~" are unobserved or latent factors, observed covariates, and/or any random effects enclosed with "<>".

Constraints

The "=" operator specifies the constraints in the model, i.e., fixing a model parameter - fixed variance, fixed covariance. The left hand side of the "=" must contain a reserved word for constraint - "var" or "cov" - with one or two variable names depending on the word. The right hand side of the "=" is a fixed value. Here are some examples:

Fixing the variance of a latent variable:
```
var(stress) = .1
```

Fixing the covariance between two latent variables:

cov(z1,z2) = 4

If a random effect such as p, e, a or c is not included in the model formulas, then, by default, the value of defaultError is included in the model. For quantitative traits, the e term should always be present. Therefore, the program automatically include e term even if no error terms are specified either in formulas nor in defaultError.

Note that "a" (additive), "p" (polygenic), "c" (common environmental) and "e"(independent environmental) are reserved variable names to specify a type of variance component, so that may not be used as input arguments.

The return value of ascertainment function is composed of two components for each pedigree, allowing either one of components or both components to be returned as a list. The first component is a TRUE/FALSE value indicating the ascertainment status of the pedigree. The second component is a vector of TRUE/FALSE stating the proband status of each member of the pedigree. Here are some examples:

Ascertainment status only

aFunction = function(ped) return(TRUE)
aFunction = function(ped) return(any(ped$SBP > 2))

The pedigree is ascertained when a "TRUE" is returned for that pedigree.

Proband status only

aFunction = function(ped) return(ped$SBP > 2) 
aFunction = function(ped) return(ped$SBP > 2 && ped$DBP < 1)
aFunction = function(ped) return(ped$disease == 1)

In this case, the ascertainment status of the pedigree is determined according to the existance of a proband in each pedigree. If no probands who meet the criteria exist in a simulated pedigree, then that pedigree is not selected.

Both

aFunction = function(ped)
{
  asc = any(ped$disease==1)    
  pro = ped$SBP > 2
  return(list(aStatus=asc, pStatus=pro))
}

When a list with both information is returned, it is required that the ascertainment status (a logical value) is the first component, and the proband status (a vector of logicals of pedigree size) is the second component. The pedigree is selected when the returned first component is a "TRUE" for that pedigree.

Examples

Run this code

## Not run: 
# # Create a strumSimModel.
# # - simForm is a formulas string.
# # - snpStrumMarker is an object of strumMarker class.
# # - aFunction is an ascertainment function definition.
# #------------------------------------------------------
# mySimModel = createSimModel(
#                  formulas = simForm,
#                  markerInfo = snpStrumMarker,
#                  ascertainment = aFunction
#                )
# ## End(Not run)