Expanding unit tests

tests/testthat/test-clustering.R

@@ -0,0 +1,318 @@
+# Tests for .cluster_gliph1() and .cluster_gliph2()
+
+# Register sequential backend for %dopar%
+foreach::registerDoSEQ()
+
+# ---- Helpers ----------------------------------------------------------------
+
+.make_cluster_data <- function() {
+  seqs <- c("CASSLAPGATNEKLFF", "CASSLDRGEVFF", "CASSYLAGGRNTLYF",
+            "CASSLTGGEETQYF", "CASSLGGRETQYF")
+  sequences <- data.frame(
+    seq_ID  = seq_along(seqs),
+    CDR3b   = seqs,
+    TRBV    = c("TRBV5-1", "TRBV6-2", "TRBV5-1", "TRBV7-2", "TRBV5-1"),
+    patient = c("P1", "P1", "P2", "P2", "P1"),
+    stringsAsFactors = FALSE
+  )
+  list(seqs = seqs, sequences = sequences)
+}
+
+# ===========================================================================
+# .cluster_gliph1() tests
+# ===========================================================================
+
+# ---- Output structure -------------------------------------------------------
+
+test_that(".cluster_gliph1 returns list with expected elements", {
+  d <- .make_cluster_data()
+
+  clone_network <- data.frame(
+    V1   = c("CASSLAPGATNEKLFF", "CASSLDRGEVFF"),
+    V2   = c("CASSLDRGEVFF", "CASSYLAGGRNTLYF"),
+    type = c("local", "local"),
+    stringsAsFactors = FALSE
+  )
+
+  result <- immGLIPH:::.cluster_gliph1(
+    clone_network   = clone_network,
+    sequences       = d$sequences,
+    not_in_global_ids = integer(0),
+    seqs            = d$seqs,
+    vgene.info      = TRUE,
+    patient.info    = TRUE,
+    global_vgene    = FALSE,
+    public_tcrs     = TRUE,
+    cluster_min_size = 1,
+    verbose         = FALSE
+  )
+
+  expect_type(result, "list")
+  expect_true("cluster_properties" %in% names(result))
+  expect_true("cluster_list" %in% names(result))
+  expect_true("clone_network" %in% names(result))
+  expect_true("save_cluster_list_df" %in% names(result))
+})
+
+# ---- Clustering logic -------------------------------------------------------
+
+test_that(".cluster_gliph1 forms connected components correctly", {
+  d <- .make_cluster_data()
+
+  # Two connected components: {seq1, seq2, seq3} and {seq4, seq5}
+  clone_network <- data.frame(
+    V1   = c("CASSLAPGATNEKLFF", "CASSLDRGEVFF",
+             "CASSLTGGEETQYF"),
+    V2   = c("CASSLDRGEVFF", "CASSYLAGGRNTLYF",
+             "CASSLGGRETQYF"),
+    type = c("local", "local", "local"),
+    stringsAsFactors = FALSE
+  )
+
+  result <- immGLIPH:::.cluster_gliph1(
+    clone_network    = clone_network,
+    sequences        = d$sequences,
+    not_in_global_ids = integer(0),
+    seqs             = d$seqs,
+    vgene.info       = TRUE,
+    patient.info     = TRUE,
+    global_vgene     = FALSE,
+    public_tcrs      = TRUE,
+    cluster_min_size = 1,
+    verbose          = FALSE
+  )
+
+  expect_s3_class(result$cluster_properties, "data.frame")
+  expect_true(nrow(result$cluster_properties) >= 2)
+  expect_true(all(c("cluster_size", "tag", "members") %in%
+                    colnames(result$cluster_properties)))
+})
+
+# ---- cluster_min_size -------------------------------------------------------
+
+test_that(".cluster_gliph1 filters by cluster_min_size", {
+  d <- .make_cluster_data()
+
+  # One pair (size 2) and one pair (size 2) and one isolated (size 1)
+  clone_network <- data.frame(
+    V1   = c("CASSLAPGATNEKLFF", "CASSLTGGEETQYF"),
+    V2   = c("CASSLDRGEVFF", "CASSLGGRETQYF"),
+    type = c("local", "local"),
+    stringsAsFactors = FALSE
+  )
+
+  # With min_size = 3, both pairs should be filtered out
+  result <- immGLIPH:::.cluster_gliph1(
+    clone_network    = clone_network,
+    sequences        = d$sequences,
+    not_in_global_ids = integer(0),
+    seqs             = d$seqs,
+    vgene.info       = TRUE,
+    patient.info     = TRUE,
+    global_vgene     = FALSE,
+    public_tcrs      = TRUE,
+    cluster_min_size = 3,
+    verbose          = FALSE
+  )
+
+  if (!is.null(result$cluster_properties)) {
+    expect_true(all(result$cluster_properties$cluster_size >= 3))
+  }
+})
+
+# ---- Empty edge list --------------------------------------------------------
+
+test_that(".cluster_gliph1 returns NULL for empty edge list", {
+  d <- .make_cluster_data()
+
+  result <- immGLIPH:::.cluster_gliph1(
+    clone_network    = NULL,
+    sequences        = d$sequences,
+    not_in_global_ids = integer(0),
+    seqs             = d$seqs,
+    vgene.info       = TRUE,
+    patient.info     = TRUE,
+    global_vgene     = FALSE,
+    public_tcrs      = TRUE,
+    cluster_min_size = 2,
+    verbose          = FALSE
+  )
+
+  expect_null(result$cluster_properties)
+  expect_equal(length(result$cluster_list), 0)
+})
+
+# ---- Singleton handling -----------------------------------------------------
+
+test_that(".cluster_gliph1 adds singletons from not_in_global_ids", {
+  d <- .make_cluster_data()
+
+  clone_network <- data.frame(
+    V1   = "CASSLAPGATNEKLFF",
+    V2   = "CASSLDRGEVFF",
+    type = "local",
+    stringsAsFactors = FALSE
+  )
+
+  # Sequence at index 5 is not in any edge
+  result <- immGLIPH:::.cluster_gliph1(
+    clone_network    = clone_network,
+    sequences        = d$sequences,
+    not_in_global_ids = c(4L, 5L),  # indices not in global edges
+    seqs             = d$seqs,
+    vgene.info       = TRUE,
+    patient.info     = TRUE,
+    global_vgene     = FALSE,
+    public_tcrs      = TRUE,
+    cluster_min_size = 1,
+    verbose          = FALSE
+  )
+
+  # The clone_network should contain singleton rows
+  expect_true(any(result$clone_network$type == "singleton"))
+})
+
+# ===========================================================================
+# .cluster_gliph2() tests
+# ===========================================================================
+
+test_that(".cluster_gliph2 returns list with expected elements", {
+  d <- .make_cluster_data()
+
+  local_res <- data.frame(
+    motif         = c("SL"),
+    start         = c(1),
+    stop          = c(2),
+    num_in_sample = c(3),
+    num_in_ref    = c(10),
+    num_fold      = c(5.0),
+    fisher.score  = c(0.001),
+    members       = paste(d$seqs[1:3], collapse = " "),
+    stringsAsFactors = FALSE
+  )
+
+  result <- immGLIPH:::.cluster_gliph2(
+    local_res              = local_res,
+    global_res             = NULL,
+    sequences              = d$sequences,
+    local_similarities     = TRUE,
+    global_similarities    = FALSE,
+    global_vgene           = FALSE,
+    all_aa_interchangeable = TRUE,
+    structboundaries       = TRUE,
+    boundary_size          = 3,
+    motif_distance_cutoff  = 1,
+    cluster_min_size       = 1,
+    boost_local_significance = FALSE,
+    verbose                = FALSE
+  )
+
+  expect_type(result, "list")
+  expect_true("merged_clusters" %in% names(result))
+  expect_true("cluster_list" %in% names(result))
+  expect_true("clone_network" %in% names(result))
+  expect_true("save_cluster_list_df" %in% names(result))
+})
+
+test_that(".cluster_gliph2 clone_network has expected edge types", {
+  d <- .make_cluster_data()
+
+  local_res <- data.frame(
+    motif         = c("SL"),
+    start         = c(1),
+    stop          = c(2),
+    num_in_sample = c(3),
+    num_in_ref    = c(10),
+    num_fold      = c(5.0),
+    fisher.score  = c(0.001),
+    members       = paste(d$seqs[1:3], collapse = " "),
+    stringsAsFactors = FALSE
+  )
+
+  result <- immGLIPH:::.cluster_gliph2(
+    local_res              = local_res,
+    global_res             = NULL,
+    sequences              = d$sequences,
+    local_similarities     = TRUE,
+    global_similarities    = FALSE,
+    global_vgene           = FALSE,
+    all_aa_interchangeable = TRUE,
+    structboundaries       = TRUE,
+    boundary_size          = 3,
+    motif_distance_cutoff  = 10,
+    cluster_min_size       = 1,
+    boost_local_significance = FALSE,
+    verbose                = FALSE
+  )
+
+  if (!is.null(result$clone_network)) {
+    expect_true("type" %in% colnames(result$clone_network))
+    # Non-NA edge types should be local, global, or singleton
+    type_col <- result$clone_network$type[!is.na(result$clone_network$type)]
+    expect_true(all(type_col %in% c("local", "global", "singleton")))
+    # At least some edges should be present
+    expect_true(length(type_col) > 0)
+  }
+})
+
+test_that(".cluster_gliph2 cluster_min_size filters small clusters", {
+  d <- .make_cluster_data()
+
+  local_res <- data.frame(
+    motif         = c("SL"),
+    start         = c(1),
+    stop          = c(2),
+    num_in_sample = c(2),
+    num_in_ref    = c(10),
+    num_fold      = c(5.0),
+    fisher.score  = c(0.001),
+    members       = paste(d$seqs[1:2], collapse = " "),
+    stringsAsFactors = FALSE
+  )
+
+  result <- immGLIPH:::.cluster_gliph2(
+    local_res              = local_res,
+    global_res             = NULL,
+    sequences              = d$sequences,
+    local_similarities     = TRUE,
+    global_similarities    = FALSE,
+    global_vgene           = FALSE,
+    all_aa_interchangeable = TRUE,
+    structboundaries       = TRUE,
+    boundary_size          = 3,
+    motif_distance_cutoff  = 10,
+    cluster_min_size       = 5,
+    boost_local_significance = FALSE,
+    verbose                = FALSE
+  )
+
+  # Cluster of size 2 should be eliminated by min_size = 5
+  if (!is.null(result$merged_clusters)) {
+    expect_true(all(result$merged_clusters$cluster_size >= 5))
+  } else {
+    expect_null(result$merged_clusters)
+  }
+})
+
+test_that(".cluster_gliph2 handles no local and no global similarities", {
+  d <- .make_cluster_data()
+
+  result <- immGLIPH:::.cluster_gliph2(
+    local_res              = NULL,
+    global_res             = NULL,
+    sequences              = d$sequences,
+    local_similarities     = FALSE,
+    global_similarities    = FALSE,
+    global_vgene           = FALSE,
+    all_aa_interchangeable = TRUE,
+    structboundaries       = TRUE,
+    boundary_size          = 3,
+    motif_distance_cutoff  = 1,
+    cluster_min_size       = 2,
+    boost_local_significance = FALSE,
+    verbose                = FALSE
+  )
+
+  expect_null(result$merged_clusters)
+  expect_equal(length(result$cluster_list), 0)
+})