tests/testthat/test-ancillary.R

In joineR: Joint Modelling of Repeated Measurements and Time-to-Event Data

library(joineR)
context("Ancillary functions")


test_that("simulation", {
  # simulate data
  set.seed(1)
  d1 <- simjoint(10, sepassoc = TRUE)
  d2 <- simjoint(10, model = "int", gamma = 1)
  d3 <- simjoint(10, model = "quad", sepassoc = TRUE, gamma = 1:3)
  # tests
  expect_output(str(d1), "List of 2")
  expect_output(str(d2), "List of 2")
  expect_output(str(d3), "List of 2")
  expect_equal(dim(d1$longitudinal)[2], 7)
  expect_equal(dim(d2$longitudinal)[2], 7)
  expect_equal(dim(d3$longitudinal)[2], 7)
  expect_equal(dim(d1$survival), c(10, 5))
  expect_equal(dim(d2$survival), c(10, 5))
  expect_equal(dim(d3$survival), c(10, 5))
  expect_warning(simjoint(10, model = "intslop", sigu = 1))
})


test_that("variogram estimation", {
  # load data + fit model
  data(mental)
  mental.unbalanced <- to.unbalanced(
    mental, id.col = 1, 
    times = c(0, 1, 2, 4, 6, 8),
    Y.col = 2:7, 
    other.col = c(8, 10, 11))
  names(mental.unbalanced)[3] <- "Y"
  vgm <- variogram(
    indv = tail(mental.unbalanced[, 1], 30),
    time = tail(mental.unbalanced[, 2], 30),
    Y = tail(mental.unbalanced[, 3], 30))
  # tests
  expect_output(str(vgm), "List of 2")
  expect_equal(dim(vgm$svar), c(75, 2))
  expect_equal(vgm$sigma2, 104, tol = 1e-03)
})


test_that("conversion between balanced + unbalanced works", {
  # convert
  simul0 <- data.frame(
    num = 1:10,
    Y1.1 = rnorm(10), Y1.2 = rnorm(10),
    Y2.1 = rnorm(10), Y2.2 = rnorm(10),
    age = rnorm(10))
  simul1 <- to.unbalanced(
    simul0, 
    id.col = 1, 
    times = c(1, 2), 
    Y.col = 2:5, 
    other.col = 6)
  simul2 <- to.balanced(
    simul1, 
    id.col = "num", 
    time.col = "time",
    Y.col = c("Y1.1", "Y2.1"), 
    other.col = "age")
  # tests
  expect_equal(simul0, simul2, check.attributes = FALSE)
  expect_equal(dim(simul0), c(10, 6))
  expect_equal(dim(simul1), c(20, 5))
  expect_output(str(simul1), "data.frame")
  expect_output(str(simul2), "data.frame")
})


test_that("subset of jointdata", {
  # data + subset
  data(heart.valve)
  heart.surv <- UniqueVariables(heart.valve,
                                var.col = c("fuyrs", "status"),
                                id.col = "num")
  heart.long <- heart.valve[, c(1, 4, 5, 7, 8, 9, 10, 11)]
  heart.jd <- jointdata(longitudinal = heart.long, 
                        survival = heart.surv,
                        id.col = "num",
                        time.col = "time")
  take <- heart.jd$survival$num[heart.jd$survival$status == 0]
  heart.jd.cens <- subset(heart.jd, take)
  # tests
  expect_output(str(heart.jd.cens), "List of 6")
  expect_output(str(heart.jd.cens$longitudinal), "data.frame")
  expect_equal(dim(heart.jd.cens$longitudinal), c(827, 8))
  expect_equal(dim(heart.jd.cens$survival), c(202, 3))
  expect_equal(heart.jd.cens$time.col, "time")
  expect_equal(heart.jd.cens$subj.col, "num")
})
  
  
test_that("jointdata without survival or baseline data", {
  # data + subset
  data(heart.valve)
  heart.long <- heart.valve[, c(1, 4, 5, 7, 8, 9, 10, 11)]
  heart.jd <- jointdata(longitudinal = heart.long, 
                        id.col = "num",
                        time.col = "time")
  # tests
  expect_output(str(heart.jd), "List of 6")
  expect_identical(heart.jd$survival, NA)
  expect_identical(heart.jd$baseline, NA)
})