Skip to contents

Leave-One-Host-Out Analysis

Source: R/loo_host_analysis.R
loo_host_analysis.Rd

Reruns codiv() with each host removed in turn and compares each run to the full analysis, showing how much each host contributes to the codiversification signal.

Usage

loo_host_analysis(
  Host_tree,
  Symbiont_tree,
  Host_to_Symbiont_df,
  hosts_to_test = NULL,
  alpha = 0.05,
  verbose = TRUE,
  ...
)

Arguments

Host_tree

A phylo object; binary phylogenetic tree of hosts

Symbiont_tree

A phylo object; binary phylogenetic tree of symbionts

Host_to_Symbiont_df

A data frame with columns "Host" and "Symbiont" linking symbionts to hosts

hosts_to_test

Character vector of host names to test. If NULL (default), tests all hosts. Use a subset to reduce computation time.

alpha

Significance threshold for counting significant nodes; default 0.05

verbose

If TRUE (default), report progress across hosts

...

Additional arguments passed to codiv() (e.g. min_hosts, min_symbiont_tips, span_fraction, permutations, methods, cores, seed)

Value

A list with elements:

  • full_results: the codiv() result on the complete data

  • loo_results: named list of the codiv() result with each host removed

  • host_importance: one row per tested host (ordered by impact_score), with the number of nodes scanned, per-method significant-node counts and their change from the full analysis (<method>_delta_n_sig), and an impact_score (mean absolute change in significant-node count across methods when the host is removed)

Details

Each run uses its own isolated checkpoint file so removing a host genuinely re-computes the affected nodes. Tree-distance metrics are disabled for speed. Runtime is roughly baseline_time x length(hosts_to_test), so consider hosts_to_test, fewer permutations, or a single method for a first pass, and cores to parallelize each run.

Examples

# \donttest{
sim <- simulate_codiv_data(n_hosts = 8, n_clades = 2, seed = 1)
loo <- loo_host_analysis(sim$host_tree, sim$symbiont_tree, sim$links,
                         permutations = 99, methods = "hommola",
                         verbose = FALSE)
loo$host_importance
#>   host n_nodes hommola_n_sig hommola_delta_n_sig hommola_mean_stat impact_score
#> 1   H8       1             0                   0        -0.5240938            0
#> 2   H1       1             0                   0        -0.4636440            0
#> 3   H6       2             0                   0         0.5264145            0
#> 4   H4       1             0                   0         0.9987138            0
#> 5   H3       1             0                   0        -0.4120341            0
#> 6   H7       2             0                   0         0.5264145            0
#> 7   H5       2             0                   0         0.5264145            0
#> 8   H2       2             0                   0         0.5264145            0
# }