visualize_subgraph_structure: Interactive association subgraphs visualization

In nstrayer/entropynet: Investigate association networks using relative association and intepretable subgraph clustering

Description

Produces an interactive plot of network structure for all possible subgraph arrangements for passed association pairs with summary statistics to guide exploration of cut-points.

Usage

visualize_subgraph_structure(
  association_pairs,
  node_info,
  subgraph_results,
  trim_subgraph_results = TRUE,
  warn_of_mismatches = TRUE,
  width = "100%",
  height = "800px",
  default_step = "min_max_rule"
)

Arguments

association_pairs	dataframe with columns a and b representing the ids of the variables or nodes and columns strength that is a numeric indicator of strength of association (higher = stronger).
node_info	Optional dataframe that has a column id that corresponds to the variables codded in a and b of association_pairs that contains additional info nade available on hover in visualization.
subgraph_results	Dataframe of subgraph results as returned by calculate_subgraph_structure. If it isnt provided it is calculated. Automatic calculation will slow down code depending on how large dataset is.
trim_subgraph_results	Should subgraph results after a giant subgraph has taken over be filtered out? Rule for filtering is at least 10% of the variables are in subgraphs and largest subgraph contains less than 95% of all variables in subgraphs. Allows for easier investigating of the subgraph structure over strength.
warn_of_mismatches	If there are differences in the ids present in association_pairs and node_info should a warning be issued?
width, height	Valid css units for output size (e.g. pixels (px) or percent(%)).
default_step	How should the default starting position of the visualization be decided. Options are min_max_rule which tends to result in lots of small subgraphs that are easier to parse, giant_component_local which tries to detect when a giant component first forms based on seen number of nodes. giant_component_global uses same logic as giant_component_local just the threshold for a giant component is based on all possible nodes rather than those seen. This gives the most accurate clusterings if true cluster structure is non-hierarchical. Alternatively you can provided the step as an integer value.

Value

Interactive javascript visualization of association network subgraphs at all possible cut-points

Examples

node_info <- dplyr::rename(virus_host_viruses, id = virus_id)
node_info$color <- ifelse(node_info$type == "RNA", "orangered", "steelblue")
visualize_subgraph_structure(
  virus_net,
  node_info = node_info
)

nstrayer/entropynet documentation built on Oct. 12, 2020, 2:20 a.m.