Prevents removing only common subsumer with resnik_similarity

Fixes #239

R/semsim.R

@@ -253,6 +253,10 @@ resnik_similarity <- function(subsumer_mat = NA, terms = NULL, ...,
     if (any(rowsToRemove)) {
       wt <- wt[! rowsToRemove]
       subsumer_mat <- subsumer_mat[! rowsToRemove,]
+      jaccard_sim <- jaccard_similarity(subsumer_mat)
+      if (any(is.nan(jaccard_sim) || jaccard_sim == 0)) {
+        stop("Subsumer matrix contained an only common subsumer between two terms with frequency of zero.", call. = FALSE)
+      }
     }
     # we assume we got frequencies, turn into IC
     wt <- -log(wt, base = base)

tests/testthat/test-semsim.R

@@ -72,6 +72,16 @@ test_that("Resnik similarity", {
   testthat::expect_equivalent(diag(sm.ic), termICs)
 })
 
+test_that("Resnik similarity prevents removing only common subsumers", {
+  # term frequency function that results in 0 frequencing for all terms removing common subsumers
+  term_freq_zero <- function(x) { rep(0, length(x))}
+  phens <- get_phenotypes("maxilla", taxon = "Cyprinidae")
+  subs.mat <- subsumer_matrix(phens$id, .colnames = "label", .labels = phens$label,
+                              preserveOrder = TRUE)
+  expect_error(resnik_similarity(subs.mat, wt=term_freq_zero),
+               "Subsumer matrix contained an only common subsumer between two terms with frequency of zero.")
+})
+
 test_that("profile similarity with Jaccard", {
   tl <- c("pelvic fin", "pectoral fin", "forelimb", "hindlimb", "dorsal fin", "caudal fin")
   tt <- sapply(tl, get_term_iri, as = "anatomy", exactOnly = TRUE)