Filter the calculated similarities — filter

This function calculates the critical value, see calculate_critical_value() and filters subsequently all the term pairs to the ones with a value ≥ critical value.

Usage

filter_similarities(
  object,
  alpha,
  permutations = 100000L,
  seed = 10101L,
  .verbose = TRUE
)

Arguments

object: OntologySim class. See OntologySim().
alpha: Float. The alpha value. For example, 0.001 would mean that the critical value is smaller than 0.1 percentile of the random permutations.
permutations: Number of random permutations.
seed: Integer. For reproducibility purposes.
.verbose: Boolean. Controls verbosity of the function.

Value

The class with filtered results added to the respective slot.

Examples

# keep only term pairs above the permutation-derived critical value
onto <- data.table::data.table(
  parent = c("a", "b", "b", "b", "c"),
  child = c("b", "c", "d", "e", "f"),
  type = c("part_of", "part_of", "part_of", "is_a", "is_a")
)
onto_obj <- OntologySim(onto, .verbose = FALSE)
onto_obj <- calculate_wang_sim_onto(
  onto_obj,
  weights = c(part_of = 0.8, is_a = 0.6),
  .verbose = FALSE
)
onto_obj <- filter_similarities(onto_obj, alpha = 0.1, .verbose = FALSE)
head(get_results(onto_obj))
#>        t1     t2       sim
#>    <char> <char>     <num>
#> 1:      f      c 0.7960848
#> 2:      d      b 0.7641509
#> 3:      b      c 0.7641509