Generate an RBH graph.

This function will generate an RBH graph based on set similarity between gene modules. You have the option to use an overlap coefficient instead of Jaccard similarity and to specify a minimum similarity.

Usage

generate_rbh_graph(
  object,
  minimum_similarity,
  k_best = 1L,
  overlap_coefficient = FALSE,
  spearman = FALSE
)

Arguments

object

The underlying class, see RbhGraph().

minimum_similarity

The minimum similarity to create an edge.

k_best

Integer. Number of best neighbours to consider. If set to 1L, this behaves as the traditional reciprocal best hit. If you set this to 3L you consider edges if the modules is in the top 3 best modules by similarity for each other.

overlap_coefficient

Boolean. Shall the overlap coefficient be used instead of Jaccard similarity. Only relevant if the underlying class is set to set similarity.

spearman

Boolean. Shall Spearman correlation be used. Only relevant if the underlying class is set to correlation-based similarity.

Value

The class with added properties.

Examples

# Jaccard-based reciprocal best hits between two module sets
set.seed(123)
modules <- data.table::data.table(
  origin = rep(c("set_a", "set_b"), each = 20),
  module = rep(c("m1", "m2", "m3", "m4"), each = 10),
  gene = unlist(replicate(4, sample(letters, 10), simplify = FALSE))
)
object <- RbhGraph(
  modules,
  rbh_type = "set",
  dataset_col = "origin",
  module_col = "module",
  value_col = "gene"
)
object <- generate_rbh_graph(object, minimum_similarity = 0)
head(get_rbh_res(object))
#>    origin target origin_modules target_modules similiarity combined_origin
#>    <char> <char>         <char>         <char>       <num>          <char>
#> 1:  set_a  set_b             m1             m3   0.2500000        set_a_m1
#> 2:  set_a  set_b             m2             m4   0.3333333        set_a_m2
#>    combined_target
#>             <char>
#> 1:        set_b_m3
#> 2:        set_b_m4