tests/testthat/test-moments_NB_PLN.R
In csoneson/compcodeR: RNAseq data simulation, differential expression analysis and performance comparison of differential expression methods
context("Moments of the Negative Binomial and Poisson Log Normal are the same")
test_that("Moments of the Negative Binomial and Log Normal are the same", {
tdir <- tempdir()
## Tree and conds
n <- 20
tree <- ape::stree(n)
tree$edge.length <- rep(10, nrow(tree$edge))
id_species <- tree$tip.label
id_species <- factor(id_species)
names(id_species) <- tree$tip.label
## Condn alternate tips
species_names <- tree$tip.label
cond_species <- rep(c(1, 2), length(species_names) / 2)
names(cond_species) <- species_names
id_cond <- id_species
id_cond <- cond_species[as.vector(id_cond)]
id_cond <- as.factor(id_cond)
## Parameters
n.vars <- 1000
relmean <- 100
reldisp <- 0.1
seqdepth <- 10000
true_mean <- seqdepth / n.vars
true_var <- reldisp * true_mean^2 + true_mean
true_sd <- sqrt(true_var)
## No tree
set.seed(18570823)
dataset <- "no_tree"
no_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = 0,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = 1,
output.file = file.path(tdir, "tmp.rds"),
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## us tree
set.seed(18570823)
dataset <- "tree"
us_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = 0,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = 1,
output.file = file.path(tdir, "tmp.rds"),
tree = tree,
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## Comparisons
mean_no_tree <- mean(no_tree_sim@count.matrix)
sd_no_tree <- sd(no_tree_sim@count.matrix)
mean_us_tree <- mean(us_tree_sim@count.matrix)
sd_us_tree <- sd(us_tree_sim@count.matrix)
testthat::expect_equal(mean_no_tree, true_mean, tol = 1e-2)
testthat::expect_equal(mean_us_tree, true_mean, tol = 1e-2)
testthat::expect_equal(sd_no_tree, true_sd, tol = 1e-2)
testthat::expect_equal(sd_us_tree, true_sd, tol = 1e-2)
## Correlation Comparisons
corup <- function(x, ...) {
cc <- cor(x, ...)
return(cc[upper.tri(cc)])
}
# Pearson correlations are equal
cor_no_tree <- corup(no_tree_sim@count.matrix)
cor_tree <- corup(us_tree_sim@count.matrix)
testthat::expect_equal(median(cor_no_tree), median(cor_tree), tol = 1e-2)
# Spearman correlations are equal
cor_no_tree <- corup(no_tree_sim@count.matrix, method = "spearman")
cor_tree <- corup(us_tree_sim@count.matrix, method = "spearman")
testthat::expect_equal(median(cor_no_tree), median(cor_tree), tol = 1e-2)
})
test_that("Moments of the Negative Binomial and Log Normal are the same, with shift", {
tdir <- tempdir()
## Tree and conds
n <- 20
tree <- ape::stree(n)
tree$edge.length <- rep(10, nrow(tree$edge))
id_species <- tree$tip.label
id_species <- factor(id_species)
names(id_species) <- tree$tip.label
## Condn alternate tips
species_names <- tree$tip.label
cond_species <- rep(c(1, 2), length(species_names) / 2)
names(cond_species) <- species_names
id_cond <- id_species
id_cond <- cond_species[as.vector(id_cond)]
id_cond <- as.factor(id_cond)
## Parameters
n.vars <- 5000
relmean <- 100
reldisp <- 0.1
seqdepth <- 50000
n.diffexp <- n.vars / 2
effect.size <- 10
true_mean_1 <- seqdepth / n.vars
true_mean_2 <- 2 * seqdepth * (effect.size + 1) / n.vars / (effect.size + 2)
true_var_1 <- reldisp * true_mean_1^2 + true_mean_1
true_sd_1 <- sqrt(true_var_1)
true_var_2 <- reldisp * true_mean_2^2 + true_mean_2
true_sd_2 <- sqrt(true_var_2)
## No tree
set.seed(18570823)
dataset <- "no_tree"
no_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## us tree
set.seed(18570823)
dataset <- "tree"
us_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
tree = tree,
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## Comparisons
mean_no_tree <- colMeans(no_tree_sim@count.matrix[1:n.diffexp, ])
sd_no_tree <- matrixStats::colSds(no_tree_sim@count.matrix[1:n.diffexp, ])
true_no_tree <- no_tree_sim@variable.annotations$truemeans.S2[1:n.diffexp] / sum(no_tree_sim@variable.annotations$truemeans.S2) * seqdepth
mean_real_no_tree <- mean(true_no_tree)
sd_real_no_tree <- sd(true_no_tree)
mean_us_tree <- colMeans(us_tree_sim@count.matrix[1:n.diffexp, ])
sd_us_tree <- matrixStats::colSds(us_tree_sim@count.matrix[1:n.diffexp, ])
true_us_tree <- us_tree_sim@variable.annotations$truemeans.S2[1:n.diffexp] / sum(us_tree_sim@variable.annotations$truemeans.S2) * seqdepth
mean_real_us_tree <- mean(true_us_tree)
sd_real_us_tree <- sd(true_us_tree)
testthat::expect_equal(mean(mean_no_tree[id_cond == 1]), true_mean_1, tol = 1e-2)
testthat::expect_equal(mean(mean_no_tree[id_cond == 2]), true_mean_2, tol = 1e-2)
testthat::expect_equal(mean(mean_no_tree[id_cond == 2]), mean_real_no_tree, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 1]), true_mean_1, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 2]), true_mean_2, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 2]), mean_real_no_tree, tol = 1e-2)
testthat::expect_equal(mean(sd_no_tree[id_cond == 1]), true_sd_1, tol = 1e-2)
testthat::expect_equal(mean(sd_us_tree[id_cond == 1]), true_sd_1, tol = 1e-2)
testthat::expect_equal(mean(sd_no_tree[id_cond == 2]),
mean(sd_us_tree[id_cond == 2]), tol = 1e-2)
## Correlation Comparisons
corup <- function(x, ...) {
cc <- cor(x, ...)
return(cc[upper.tri(cc)])
}
# Pearson correlations are equal
cor_no_tree <- corup(no_tree_sim@count.matrix)
cor_tree <- corup(us_tree_sim@count.matrix)
testthat::expect_equal(median(cor_no_tree), median(cor_tree), tol = 1e-2)
# Spearman correlations are equal
cor_no_tree <- corup(no_tree_sim@count.matrix, method = "spearman")
cor_tree <- corup(us_tree_sim@count.matrix, method = "spearman")
testthat::expect_equal(median(cor_no_tree), median(cor_tree), tol = 1e-2)
})
test_that("Moments of the Negative Binomial and Log Normal are the same, varying mean", {
tdir <- tempdir()
## Tree and conds
n <- 14
tree <- ape::stree(n)
tree$edge.length <- rep(1, nrow(tree$edge))
id_species <- tree$tip.label
id_species <- factor(id_species)
tree$tip.label <- paste0(tree$tip.label, "_rep")
names(id_species) <- tree$tip.label
## Condn alternate tips
species_names <- tree$tip.label
cond_species <- c(2, 1, 2, 1, 2, 2, 1, 1, 2, 2, 1, 1, 1, 1) #rep(c(1, 2), length(species_names) / 2)
names(cond_species) <- species_names
id_cond <- cond_species
id_cond <- as.factor(id_cond)
effect.size <- 0
## Parameters
set.seed(18570823)
n.vars <- 3560
length.mu.phi.estimates <- readRDS(system.file("extdata", "Stern2018.Length.Mu.Phi.Estimates.rds", package = "compcodeR"))
relmean <- length.mu.phi.estimates$stern2018.length.mu
reldisp <- length.mu.phi.estimates$stern2018.length.phi
seqdepth <- 2000000
n.diffexp <- 0
nrep <- 1
## No tree
set.seed(18570821)
dataset <- "no_tree"
count_no_tree <- array(NA, dim = c(n.vars, n, nrep))
for (rep in 1:nrep) {
no_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = relmean,
dispersions = reldisp,
fraction.upregulated = 0.5,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
# tree = tree,
# model.process = "BM",
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
count_no_tree[, , rep] <- no_tree_sim@count.matrix
}
## us tree
set.seed(18570821)
dataset <- "tree"
count_tree <- array(NA, dim = c(n.vars, n, nrep))
for (rep in 1:nrep) {
us_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = relmean,
dispersions = reldisp,
fraction.upregulated = 0.5,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
tree = tree,
model.process = "BM",
prop.var.tree = 1.0,
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
count_tree[, , rep] <- us_tree_sim@count.matrix
}
## Correlation Comparisons
corup <- function(x, ...) {
cc <- cor(x, ...)
return(cc[upper.tri(cc)])
}
# Pearson correlations are equal
cor_no_tree <- rowMeans(apply(count_no_tree, 3, corup))
cor_tree <- rowMeans(apply(count_tree, 3, corup))
testthat::expect_equal(median(cor_no_tree), median(cor_tree), tol = 1e-2)
# But Spearman are not correlations not are equal
cor_no_tree <- rowMeans(apply(count_no_tree, 3, corup, method = "spearman"))
cor_tree <- rowMeans(apply(count_tree, 3, corup, method = "spearman"))
testthat::expect_true(median(cor_no_tree) <= median(cor_tree))
})
test_that("Moments of the Negative Binomial and Log Normal are the same, with shift, random tree", {
tdir <- tempdir()
## Tree and conds
n <- 20
set.seed(1789)
tree <- tree <- ape::rphylo(n, 1, 0)
id_species <- tree$tip.label
id_species <- factor(id_species)
names(id_species) <- tree$tip.label
## Condn alternate tips
species_names <- tree$tip.label
cond_species <- rep(c(1, 2), length(species_names) / 2)
names(cond_species) <- species_names
id_cond <- id_species
id_cond <- cond_species[as.vector(id_cond)]
id_cond <- as.factor(id_cond)
## Parameters
n.vars <- 5000
relmean <- 100
reldisp <- 0.1
seqdepth <- 50000
n.diffexp <- n.vars / 2
effect.size <- 10
true_mean_1 <- seqdepth / n.vars
true_mean_2 <- 2 * seqdepth * (effect.size + 1) / n.vars / (effect.size + 2)
true_var_1 <- reldisp * true_mean_1^2 + true_mean_1
true_sd_1 <- sqrt(true_var_1)
true_var_2 <- reldisp * true_mean_2^2 + true_mean_2
true_sd_2 <- sqrt(true_var_2)
## No tree
set.seed(18570823)
dataset <- "no_tree"
no_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## us tree
set.seed(18570823)
dataset <- "tree"
us_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
tree = tree,
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## Comparisons
mean_no_tree <- colMeans(no_tree_sim@count.matrix[1:n.diffexp, ])
sd_no_tree <- matrixStats::colSds(no_tree_sim@count.matrix[1:n.diffexp, ])
mean_us_tree <- colMeans(us_tree_sim@count.matrix[1:n.diffexp, ])
sd_us_tree <- matrixStats::colSds(us_tree_sim@count.matrix[1:n.diffexp, ])
testthat::expect_equal(mean(mean_no_tree[id_cond == 1]), true_mean_1, tol = 1e-2)
testthat::expect_equal(mean(mean_no_tree[id_cond == 2]), true_mean_2, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 1]), true_mean_1, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 2]), true_mean_2, tol = 1e-2)
testthat::expect_equal(mean(sd_no_tree[id_cond == 1]), true_sd_1, tol = 1e-2)
testthat::expect_equal(mean(sd_us_tree[id_cond == 1]), true_sd_1, tol = 1e-2)
testthat::expect_equal(mean(sd_no_tree[id_cond == 2]),
mean(sd_us_tree[id_cond == 2]), tol = 1e-2)
})
test_that("Moments of the Negative Binomial and Log Normal are the same, with shift, random tree, with replicates and varying prop.var.tree", {
skip_if_not_installed("phytools")
tdir <- tempdir()
## Tree and conds
n <- 20
set.seed(1789)
tree_norep <- ape::rphylo(n, 1, 0)
r <- 2
n <- r * n
tree <- add_replicates(tree_norep, r)
id_species <- sub("_.*$", "", tree$tip.label)
id_species <- factor(id_species)
names(id_species) <- tree$tip.label
## Condn alternate tips
species_names <- unique(sub("_.*$", "", tree$tip.label))
cond_species <- rep(c(1, 2), length(species_names) / 2)
names(cond_species) <- species_names
id_cond <- id_species
id_cond <- cond_species[as.vector(id_cond)]
id_cond <- as.factor(id_cond)
names(id_cond) <- names(id_species)
## Parameters
n.vars <- 5000
relmean <- 100
reldisp <- 0.1
seqdepth <- 50000
n.diffexp <- n.vars / 2
effect.size <- 2
true_mean_1 <- seqdepth / n.vars
true_mean_2 <- 2 * seqdepth * (effect.size + 1) / n.vars / (effect.size + 2)
true_var_1 <- reldisp * true_mean_1^2 + true_mean_1
true_sd_1 <- sqrt(true_var_1)
true_var_2 <- reldisp * true_mean_2^2 + true_mean_2
true_sd_2 <- sqrt(true_var_2)
## No tree
set.seed(18570823)
dataset <- "no_tree"
no_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## us tree
set.seed(18570823)
dataset <- "tree"
us_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
tree = tree,
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE,
prop.var.tree = runif(n.vars)
)
## Comparisons
mean_no_tree <- colMeans(no_tree_sim@count.matrix[1:n.diffexp, ])
sd_no_tree <- matrixStats::colSds(no_tree_sim@count.matrix[1:n.diffexp, ])
mean_us_tree <- colMeans(us_tree_sim@count.matrix[1:n.diffexp, ])
sd_us_tree <- matrixStats::colSds(us_tree_sim@count.matrix[1:n.diffexp, ])
testthat::expect_equal(mean(mean_no_tree[id_cond == 1]), true_mean_1, tol = 1e-2)
testthat::expect_equal(mean(mean_no_tree[id_cond == 2]), true_mean_2, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 1]), true_mean_1, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 2]), true_mean_2, tol = 1e-2)
testthat::expect_equal(mean(sd_no_tree[id_cond == 1]), true_sd_1, tol = 1e-2)
testthat::expect_equal(mean(sd_us_tree[id_cond == 1]), true_sd_1, tol = 1e-2)
testthat::expect_equal(mean(sd_no_tree[id_cond == 2]),
mean(sd_us_tree[id_cond == 2]), tol = 1e-2)
})
csoneson/compcodeR documentation built on Dec. 23, 2024, 10:42 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
context("Moments of the Negative Binomial and Poisson Log Normal are the same")
test_that("Moments of the Negative Binomial and Log Normal are the same", {
tdir <- tempdir()
## Tree and conds
n <- 20
tree <- ape::stree(n)
tree$edge.length <- rep(10, nrow(tree$edge))
id_species <- tree$tip.label
id_species <- factor(id_species)
names(id_species) <- tree$tip.label
## Condn alternate tips
species_names <- tree$tip.label
cond_species <- rep(c(1, 2), length(species_names) / 2)
names(cond_species) <- species_names
id_cond <- id_species
id_cond <- cond_species[as.vector(id_cond)]
id_cond <- as.factor(id_cond)
## Parameters
n.vars <- 1000
relmean <- 100
reldisp <- 0.1
seqdepth <- 10000
true_mean <- seqdepth / n.vars
true_var <- reldisp * true_mean^2 + true_mean
true_sd <- sqrt(true_var)
## No tree
set.seed(18570823)
dataset <- "no_tree"
no_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = 0,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = 1,
output.file = file.path(tdir, "tmp.rds"),
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## us tree
set.seed(18570823)
dataset <- "tree"
us_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = 0,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = 1,
output.file = file.path(tdir, "tmp.rds"),
tree = tree,
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## Comparisons
mean_no_tree <- mean(no_tree_sim@count.matrix)
sd_no_tree <- sd(no_tree_sim@count.matrix)
mean_us_tree <- mean(us_tree_sim@count.matrix)
sd_us_tree <- sd(us_tree_sim@count.matrix)
testthat::expect_equal(mean_no_tree, true_mean, tol = 1e-2)
testthat::expect_equal(mean_us_tree, true_mean, tol = 1e-2)
testthat::expect_equal(sd_no_tree, true_sd, tol = 1e-2)
testthat::expect_equal(sd_us_tree, true_sd, tol = 1e-2)
## Correlation Comparisons
corup <- function(x, ...) {
cc <- cor(x, ...)
return(cc[upper.tri(cc)])
}
# Pearson correlations are equal
cor_no_tree <- corup(no_tree_sim@count.matrix)
cor_tree <- corup(us_tree_sim@count.matrix)
testthat::expect_equal(median(cor_no_tree), median(cor_tree), tol = 1e-2)
# Spearman correlations are equal
cor_no_tree <- corup(no_tree_sim@count.matrix, method = "spearman")
cor_tree <- corup(us_tree_sim@count.matrix, method = "spearman")
testthat::expect_equal(median(cor_no_tree), median(cor_tree), tol = 1e-2)
})
test_that("Moments of the Negative Binomial and Log Normal are the same, with shift", {
tdir <- tempdir()
## Tree and conds
n <- 20
tree <- ape::stree(n)
tree$edge.length <- rep(10, nrow(tree$edge))
id_species <- tree$tip.label
id_species <- factor(id_species)
names(id_species) <- tree$tip.label
## Condn alternate tips
species_names <- tree$tip.label
cond_species <- rep(c(1, 2), length(species_names) / 2)
names(cond_species) <- species_names
id_cond <- id_species
id_cond <- cond_species[as.vector(id_cond)]
id_cond <- as.factor(id_cond)
## Parameters
n.vars <- 5000
relmean <- 100
reldisp <- 0.1
seqdepth <- 50000
n.diffexp <- n.vars / 2
effect.size <- 10
true_mean_1 <- seqdepth / n.vars
true_mean_2 <- 2 * seqdepth * (effect.size + 1) / n.vars / (effect.size + 2)
true_var_1 <- reldisp * true_mean_1^2 + true_mean_1
true_sd_1 <- sqrt(true_var_1)
true_var_2 <- reldisp * true_mean_2^2 + true_mean_2
true_sd_2 <- sqrt(true_var_2)
## No tree
set.seed(18570823)
dataset <- "no_tree"
no_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## us tree
set.seed(18570823)
dataset <- "tree"
us_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
tree = tree,
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## Comparisons
mean_no_tree <- colMeans(no_tree_sim@count.matrix[1:n.diffexp, ])
sd_no_tree <- matrixStats::colSds(no_tree_sim@count.matrix[1:n.diffexp, ])
true_no_tree <- no_tree_sim@variable.annotations$truemeans.S2[1:n.diffexp] / sum(no_tree_sim@variable.annotations$truemeans.S2) * seqdepth
mean_real_no_tree <- mean(true_no_tree)
sd_real_no_tree <- sd(true_no_tree)
mean_us_tree <- colMeans(us_tree_sim@count.matrix[1:n.diffexp, ])
sd_us_tree <- matrixStats::colSds(us_tree_sim@count.matrix[1:n.diffexp, ])
true_us_tree <- us_tree_sim@variable.annotations$truemeans.S2[1:n.diffexp] / sum(us_tree_sim@variable.annotations$truemeans.S2) * seqdepth
mean_real_us_tree <- mean(true_us_tree)
sd_real_us_tree <- sd(true_us_tree)
testthat::expect_equal(mean(mean_no_tree[id_cond == 1]), true_mean_1, tol = 1e-2)
testthat::expect_equal(mean(mean_no_tree[id_cond == 2]), true_mean_2, tol = 1e-2)
testthat::expect_equal(mean(mean_no_tree[id_cond == 2]), mean_real_no_tree, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 1]), true_mean_1, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 2]), true_mean_2, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 2]), mean_real_no_tree, tol = 1e-2)
testthat::expect_equal(mean(sd_no_tree[id_cond == 1]), true_sd_1, tol = 1e-2)
testthat::expect_equal(mean(sd_us_tree[id_cond == 1]), true_sd_1, tol = 1e-2)
testthat::expect_equal(mean(sd_no_tree[id_cond == 2]),
mean(sd_us_tree[id_cond == 2]), tol = 1e-2)
## Correlation Comparisons
corup <- function(x, ...) {
cc <- cor(x, ...)
return(cc[upper.tri(cc)])
}
# Pearson correlations are equal
cor_no_tree <- corup(no_tree_sim@count.matrix)
cor_tree <- corup(us_tree_sim@count.matrix)
testthat::expect_equal(median(cor_no_tree), median(cor_tree), tol = 1e-2)
# Spearman correlations are equal
cor_no_tree <- corup(no_tree_sim@count.matrix, method = "spearman")
cor_tree <- corup(us_tree_sim@count.matrix, method = "spearman")
testthat::expect_equal(median(cor_no_tree), median(cor_tree), tol = 1e-2)
})
test_that("Moments of the Negative Binomial and Log Normal are the same, varying mean", {
tdir <- tempdir()
## Tree and conds
n <- 14
tree <- ape::stree(n)
tree$edge.length <- rep(1, nrow(tree$edge))
id_species <- tree$tip.label
id_species <- factor(id_species)
tree$tip.label <- paste0(tree$tip.label, "_rep")
names(id_species) <- tree$tip.label
## Condn alternate tips
species_names <- tree$tip.label
cond_species <- c(2, 1, 2, 1, 2, 2, 1, 1, 2, 2, 1, 1, 1, 1) #rep(c(1, 2), length(species_names) / 2)
names(cond_species) <- species_names
id_cond <- cond_species
id_cond <- as.factor(id_cond)
effect.size <- 0
## Parameters
set.seed(18570823)
n.vars <- 3560
length.mu.phi.estimates <- readRDS(system.file("extdata", "Stern2018.Length.Mu.Phi.Estimates.rds", package = "compcodeR"))
relmean <- length.mu.phi.estimates$stern2018.length.mu
reldisp <- length.mu.phi.estimates$stern2018.length.phi
seqdepth <- 2000000
n.diffexp <- 0
nrep <- 1
## No tree
set.seed(18570821)
dataset <- "no_tree"
count_no_tree <- array(NA, dim = c(n.vars, n, nrep))
for (rep in 1:nrep) {
no_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = relmean,
dispersions = reldisp,
fraction.upregulated = 0.5,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
# tree = tree,
# model.process = "BM",
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
count_no_tree[, , rep] <- no_tree_sim@count.matrix
}
## us tree
set.seed(18570821)
dataset <- "tree"
count_tree <- array(NA, dim = c(n.vars, n, nrep))
for (rep in 1:nrep) {
us_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = relmean,
dispersions = reldisp,
fraction.upregulated = 0.5,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
tree = tree,
model.process = "BM",
prop.var.tree = 1.0,
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
count_tree[, , rep] <- us_tree_sim@count.matrix
}
## Correlation Comparisons
corup <- function(x, ...) {
cc <- cor(x, ...)
return(cc[upper.tri(cc)])
}
# Pearson correlations are equal
cor_no_tree <- rowMeans(apply(count_no_tree, 3, corup))
cor_tree <- rowMeans(apply(count_tree, 3, corup))
testthat::expect_equal(median(cor_no_tree), median(cor_tree), tol = 1e-2)
# But Spearman are not correlations not are equal
cor_no_tree <- rowMeans(apply(count_no_tree, 3, corup, method = "spearman"))
cor_tree <- rowMeans(apply(count_tree, 3, corup, method = "spearman"))
testthat::expect_true(median(cor_no_tree) <= median(cor_tree))
})
test_that("Moments of the Negative Binomial and Log Normal are the same, with shift, random tree", {
tdir <- tempdir()
## Tree and conds
n <- 20
set.seed(1789)
tree <- tree <- ape::rphylo(n, 1, 0)
id_species <- tree$tip.label
id_species <- factor(id_species)
names(id_species) <- tree$tip.label
## Condn alternate tips
species_names <- tree$tip.label
cond_species <- rep(c(1, 2), length(species_names) / 2)
names(cond_species) <- species_names
id_cond <- id_species
id_cond <- cond_species[as.vector(id_cond)]
id_cond <- as.factor(id_cond)
## Parameters
n.vars <- 5000
relmean <- 100
reldisp <- 0.1
seqdepth <- 50000
n.diffexp <- n.vars / 2
effect.size <- 10
true_mean_1 <- seqdepth / n.vars
true_mean_2 <- 2 * seqdepth * (effect.size + 1) / n.vars / (effect.size + 2)
true_var_1 <- reldisp * true_mean_1^2 + true_mean_1
true_sd_1 <- sqrt(true_var_1)
true_var_2 <- reldisp * true_mean_2^2 + true_mean_2
true_sd_2 <- sqrt(true_var_2)
## No tree
set.seed(18570823)
dataset <- "no_tree"
no_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## us tree
set.seed(18570823)
dataset <- "tree"
us_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
tree = tree,
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## Comparisons
mean_no_tree <- colMeans(no_tree_sim@count.matrix[1:n.diffexp, ])
sd_no_tree <- matrixStats::colSds(no_tree_sim@count.matrix[1:n.diffexp, ])
mean_us_tree <- colMeans(us_tree_sim@count.matrix[1:n.diffexp, ])
sd_us_tree <- matrixStats::colSds(us_tree_sim@count.matrix[1:n.diffexp, ])
testthat::expect_equal(mean(mean_no_tree[id_cond == 1]), true_mean_1, tol = 1e-2)
testthat::expect_equal(mean(mean_no_tree[id_cond == 2]), true_mean_2, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 1]), true_mean_1, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 2]), true_mean_2, tol = 1e-2)
testthat::expect_equal(mean(sd_no_tree[id_cond == 1]), true_sd_1, tol = 1e-2)
testthat::expect_equal(mean(sd_us_tree[id_cond == 1]), true_sd_1, tol = 1e-2)
testthat::expect_equal(mean(sd_no_tree[id_cond == 2]),
mean(sd_us_tree[id_cond == 2]), tol = 1e-2)
})
test_that("Moments of the Negative Binomial and Log Normal are the same, with shift, random tree, with replicates and varying prop.var.tree", {
skip_if_not_installed("phytools")
tdir <- tempdir()
## Tree and conds
n <- 20
set.seed(1789)
tree_norep <- ape::rphylo(n, 1, 0)
r <- 2
n <- r * n
tree <- add_replicates(tree_norep, r)
id_species <- sub("_.*$", "", tree$tip.label)
id_species <- factor(id_species)
names(id_species) <- tree$tip.label
## Condn alternate tips
species_names <- unique(sub("_.*$", "", tree$tip.label))
cond_species <- rep(c(1, 2), length(species_names) / 2)
names(cond_species) <- species_names
id_cond <- id_species
id_cond <- cond_species[as.vector(id_cond)]
id_cond <- as.factor(id_cond)
names(id_cond) <- names(id_species)
## Parameters
n.vars <- 5000
relmean <- 100
reldisp <- 0.1
seqdepth <- 50000
n.diffexp <- n.vars / 2
effect.size <- 2
true_mean_1 <- seqdepth / n.vars
true_mean_2 <- 2 * seqdepth * (effect.size + 1) / n.vars / (effect.size + 2)
true_var_1 <- reldisp * true_mean_1^2 + true_mean_1
true_sd_1 <- sqrt(true_var_1)
true_var_2 <- reldisp * true_mean_2^2 + true_mean_2
true_sd_2 <- sqrt(true_var_2)
## No tree
set.seed(18570823)
dataset <- "no_tree"
no_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE
)
## us tree
set.seed(18570823)
dataset <- "tree"
us_tree_sim <- compcodeR::generateSyntheticData(dataset = dataset,
n.vars = n.vars,
samples.per.cond = n / 2,
n.diffexp = n.diffexp,
repl.id = 1,
seqdepth = seqdepth,
minfact = 1,
maxfact = 1,
relmeans = rep(relmean, n.vars),
dispersions = rep(reldisp, n.vars),
fraction.upregulated = 1,
between.group.diffdisp = FALSE,
filter.threshold.total = 1,
filter.threshold.mediancpm = 0,
fraction.non.overdispersed = 0,
random.outlier.high.prob = 0,
random.outlier.low.prob = 0,
single.outlier.high.prob = 0,
single.outlier.low.prob = 0,
effect.size = effect.size,
output.file = file.path(tdir, "tmp.rds"),
tree = tree,
id.condition = id_cond,
id.species = id_species,
lengths.relmeans = NULL,
lengths.dispersions = NULL,
lengths.phylo = FALSE,
prop.var.tree = runif(n.vars)
)
## Comparisons
mean_no_tree <- colMeans(no_tree_sim@count.matrix[1:n.diffexp, ])
sd_no_tree <- matrixStats::colSds(no_tree_sim@count.matrix[1:n.diffexp, ])
mean_us_tree <- colMeans(us_tree_sim@count.matrix[1:n.diffexp, ])
sd_us_tree <- matrixStats::colSds(us_tree_sim@count.matrix[1:n.diffexp, ])
testthat::expect_equal(mean(mean_no_tree[id_cond == 1]), true_mean_1, tol = 1e-2)
testthat::expect_equal(mean(mean_no_tree[id_cond == 2]), true_mean_2, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 1]), true_mean_1, tol = 1e-2)
testthat::expect_equal(mean(mean_us_tree[id_cond == 2]), true_mean_2, tol = 1e-2)
testthat::expect_equal(mean(sd_no_tree[id_cond == 1]), true_sd_1, tol = 1e-2)
testthat::expect_equal(mean(sd_us_tree[id_cond == 1]), true_sd_1, tol = 1e-2)
testthat::expect_equal(mean(sd_no_tree[id_cond == 2]),
mean(sd_us_tree[id_cond == 2]), tol = 1e-2)
})
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.