FamEvent-package: Family age-at-onset data simulation and penetrance estimation
In FamEvent: Family Age-at-Onset Data Simulation and Penetrance Estimation

FamEvent-package

R Documentation

Family age-at-onset data simulation and penetrance estimation

Description

Family-based studies are used to characterize the disease risk associated with being a carrier of a major gene. When the disease risk can vary with age of onset, penetrance or disease risk functions need to provide age-dependent estimates of this disease risk over lifetime. This FamEvent package can generate age-at-onset data in the context of familial studies, with correction for ascertainment (selection) bias arising from a specified study design based on proband's mutation and disease statuses. Possible study designs include: "pop" for a population-based design where families are ascertained through affected probands; "pop+", similar to "pop", but with probands also known to be mutation carriers; "cli" for a clinic-based design that includes affected probands with at least one affected parent and one affected sibling; "cli+", similar to "cli", but with probands also known to be mutation carrriers; "twostage" for a two-stage design that randomly samples families from the population in the first stage and oversamples high risk families with at least two affected members at the second stage; and "noasc" for designs with no ascertainment criteria applied.

Ages at disease onset are generated specific to family members' gender and mutation status according to the specified study design with residual familial correlations induced by a shared frailty, correlated frailties with Kinship matrix, or a second gene.

For estimating age at onset risks with family data, an ascertainment corrected prospective likelihood approach is used to account for the population or clinic-based study designs while a composite likelihood approach is used for the two-stage sampling design. The Expectation and Maximization (EM) algorithm has been implemented for inferring missing genotypes conditional on observed genotypes and phenotypes in the families. For family members who have missing genotypes, their carrier probabilities are obtained either from the fitted model or from Mendelian transmission probabilities.

This package also provides functions to plot the age-dependent penetrance curves estimated parametrically from the fitted model or non-parametrically from the data, pedigree plots of simulated families and penetrance function curves for carriers and non-carriers of a major and second gene based on specified parameter values.

In summary, this package facilitates the following:

1. Family data generations from 1) a shared frailty model or 2) a correlated frailty model with Kinship and/or IBD matrix; 3) a competing risk model.

2. Covariates considered: sex, gene, options for additional covariates.

3. Time-varying covariate generation based on permanent exposure (PE) or the Cox-Oaks model (CO).

4. Model estimation for shared frailty models and competing risk models.

5. Updates for this version: penmodel and penmodelEM can fit models with a single covariate, and accommodate families with either affected and unaffected probands.

Author(s)

Yun-Hee Choi, Karen Kopciuk, Laurent Briollais, Wenqing He

Maintainer: Yun-Hee Choi < yun-hee.choi@schulich.uwo.ca >

References

Choi, Y.-H., Jung, H., Buys, S., Daly, M., John, E.M., Hopper, J., Andrulis, I., Terry, M.B., Briollais, L. (2021) A Competing Risks Model with Binary Time Varying Covariates for Estimation of Breast Cancer Risks in BRCA1 Families, Statistical Methods in Medical Research 30 (9), 2165-2183. https://doi.org/10.1177/09622802211008945.

Choi, Y.-H., Briollais, L., He, W. and Kopciuk, K. (2021) FamEvent: An R Package for Generating and Modeling Time-to-Event Data in Family Designs, Journal of Statistical Software 97 (7), 1-30. doi:10.18637/jss.v097.i07

Choi, Y.-H., Kopciuk, K. and Briollais, L. (2008) Estimating Disease Risk Associated Mutated Genes in Family-Based Designs, Human Heredity 66, 238-251.

Choi, Y.-H. and Briollais (2011) An EM Composite Likelihood Approach for Multistage Sampling of Family Data with Missing Genetic Covariates, Statistica Sinica 21, 231-253.

Examples

## Not run: 
# Example1:  Simulate family data 
set.seed(4321)
fam <- simfam(N.fam = 100, design = "pop+", variation = "none", 
       base.dist = "Weibull", base.parms = c(0.01, 3), vbeta = c(-1.13, 2.35), 
       allelefreq = 0.02)

# Summary of simulated family data
summary(fam) 

# Plot pedigrees for family 1 and 2
plot(fam, famid = c(1,2))

# Plot penetrance functions with Weibull baseline 
penplot(base.parms = c(0.01, 3), vbeta = c(-1.3, 2.35), base.dist = "Weibull", 
        variation = "none", agemin = 20)

# Fit a penetrance model with Weibull baseline hazard to family data 
fit <- penmodel(Surv(time, status) ~ gender + mgene, cluster = "famID", 
       design = "pop+", parms = c(0.01, 3, -1.13, 2.35), data = fam, 
       base.dist = "Weibull", robust = TRUE)

# Summary of estimated model parameters and penetrance estimates 
summary(fit)

# Plot penetrance curves for model checking 
plot(fit) 

## Example 2: Simulate family data from a correlated frailty model 
#  using Kinship and IBD matrices derived from pedigree data.

# IBD matrix and inputdata must be provided.
# The dataset below is generated by simfam() for use as inputdata.

data <- simfam(N.fam = 10, design = "noasc", variation = "none",
         base.dist = "Weibull", base.parms = c(0.016, 3), vbeta = c(1, 1))

IBDmatrix <- diag(1, dim(data)[1])
data <- data[ , c(1:7, 11, 14)]

fam2 <- simfam2(inputdata = data, IBD = IBDmatrix, design = "pop", 
        variation = c("kinship","IBD"), depend = c(1, 1), 
        base.dist = "Weibull", base.parms = c(0.016, 3),
        var_names = c("gender", "mgene"), vbeta = c(1,1),
        agemin = 20) 

summary(fam2)

### Example 3: Simulate correlated competing risks family data 
set.seed(4321)
fam3 <- simfam_cmp(N.fam = 200, design = "pop+", variation = "frailty", 
       base.dist = "Weibull", frailty.dist = "cgamma", depend=c(1, 2, 0.5), 
       allelefreq = 0.02, base.parms = list(c(0.01, 3), c(0.01, 3)), 
       vbeta = list(c(-1.13, 2.35), c(-1, 2)))

# Summary of simulated family data
summary(fam3) 

# Plot the pedigree for Family 1 
plot(fam3, famid = 1)

# Plot the penetrance function for event 1 with a Weibull baseline hazard 
penplot_cmp(event = 1, base.parms = list(c(0.01, 3), c(0.01, 3)), 
          vbeta = list(c(-1.3, 2.35), c(-1, 2)), base.dist = "Weibull", 
          variation = "frailty", frailty.dist = "cgamma", 
          depend = c(1, 2, 0.5), agemin = 20)


# Fit a shared correlated gamma frailty model to simulated competing risk data

fit3 <- penmodel_cmp(
        formula1 = Surv(time, status == 1) ~ gender + mgene, 
        formula2 = Surv(time, status == 2) ~ gender + mgene,
        cluster = "famID", gvar = "mgene", design = "pop+",  
        parms = list(c(0.01, 3, -1, 2), c(0.01, 3, -1, 2), c(0.5, 1, 0.5)),
        base.dist = "Weibull", frailty.dist = "cgamma", data = fam2, 
        robust = TRUE)

# Summary of the model parameter estimates from the fitted model
summary(fit3)

# Plot the lifetime penetrance curves with 95
# specific  gender and mutation status groups, together with the corresponding 
# nonparametric penetrance curves based on data excluding probands. 

plot(fit3, add.CIF = TRUE, conf.int = TRUE, MC = 100)

### Example 4: Simulate family data with a time-varying covariate 
set.seed(4321)

fam4 <- simfam_tvc(N.fam = 10, design = "pop", variation = "frailty", 
       base.dist = "Weibull", frailty.dist = "gamma", depend = 1, 
       add.tvc = TRUE, tvc.type = "CO", tvc.range = c(30,60), 
       tvc.parms = c(1, 0.1, 0), allelefreq = 0.02, 
       base.parms = c(0.01, 3), vbeta = c(-1.13, 2.35))

summary(fam4)


## End(Not run)

FamEvent documentation built on May 15, 2026, 1:06 a.m.