# DEG Functions
library(singleCellTK)
context("Testing DEG functions")
data(sceBatches, package = "singleCellTK")
logcounts(sceBatches) <- log1p(counts(sceBatches))
sceBatches <- subsetSCECols(sceBatches, colData = "batch == 'w'")
test_that(desc = "Testing Limma DE", {
sceBatches <- runLimmaDE(inSCE = sceBatches,
class = "cell_type",
classGroup1 = "alpha", classGroup2 = "beta",
groupName1 = "a", groupName2 = "b",
analysisName = "aVSbLimma")
testthat::expect_true("diffExp" %in% names(metadata(sceBatches)))
testthat::expect_true("aVSbLimma" %in% names(metadata(sceBatches)$diffExp))
})
test_that(desc = "Testing MAST DE", {
sceBatches <- runMAST(inSCE = sceBatches,
class = "cell_type",
classGroup1 = "alpha", classGroup2 = "beta",
groupName1 = "a", groupName2 = "b",
analysisName = "aVSbMAST")
testthat::expect_true("diffExp" %in% names(metadata(sceBatches)))
testthat::expect_true("aVSbMAST" %in% names(metadata(sceBatches)$diffExp))
})
test_that(desc = "Testing DESeq2 DE", {
sceBatches <- runDESeq2(inSCE = sceBatches,
class = "cell_type",
classGroup1 = "alpha", classGroup2 = "beta",
groupName1 = "a", groupName2 = "b",
analysisName = "aVSbDESeq2")
testthat::expect_true("diffExp" %in% names(metadata(sceBatches)))
testthat::expect_true("aVSbDESeq2" %in% names(metadata(sceBatches)$diffExp))
})
test_that(desc = "Testing ANOVA DE", {
sceBatches <- runANOVA(inSCE = sceBatches,
class = "cell_type",
classGroup1 = "alpha", classGroup2 = "beta",
groupName1 = "a", groupName2 = "b",
analysisName = "aVSbANOVA")
testthat::expect_true("diffExp" %in% names(metadata(sceBatches)))
testthat::expect_true("aVSbANOVA" %in% names(metadata(sceBatches)$diffExp))
})
test_that(desc = "Testing Wilcoxon DE", {
sceBatches <- runWilcox(inSCE = sceBatches,
class = "cell_type",
classGroup1 = "alpha", classGroup2 = "beta",
groupName1 = "a", groupName2 = "b",
analysisName = "aVSbWilcox")
testthat::expect_true("diffExp" %in% names(metadata(sceBatches)))
testthat::expect_true("aVSbWilcox" %in% names(metadata(sceBatches)$diffExp))
# Also Plotting functions at this point
vlcn <- plotDEGVolcano(sceBatches, "aVSbWilcox")
testthat::expect_is(vlcn, "ggplot")
hm <- plotDEGHeatmap(sceBatches, "aVSbWilcox",
minGroup1ExprPerc = NULL, maxGroup2ExprPerc = NULL)
testthat::expect_is(hm, "Heatmap")
pR <- plotDEGRegression(sceBatches, "aVSbWilcox")
testthat::expect_is(pR, "ggplot")
pV <- plotDEGViolin(sceBatches, "aVSbWilcox")
testthat::expect_is(pV, "ggplot")
})
test_that(desc = "Testing findMarker", {
sceBatches <- runFindMarker(inSCE = sceBatches,
cluster = "cell_type")
testthat::expect_true("findMarker" %in% names(metadata(sceBatches)))
topTable <- getFindMarkerTopTable(sceBatches, log2fcThreshold = 1,
fdrThreshold = 0.05, minClustExprPerc = 0.7,
maxCtrlExprPerc = 0.4, minMeanExpr = 1,
topN = 10)
testthat::expect_is(topTable, "data.frame")
testthat::expect_named(topTable, c("Gene", "Log2_FC", "Pvalue", "FDR",
"cell_type", "clusterExprPerc",
"ControlExprPerc", "clusterAveExpr"))
testthat::expect_gt(nrow(topTable), 0)
hmFM <- plotFindMarkerHeatmap(sceBatches)
testthat::expect_is(hmFM, "Heatmap")
})
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.