6  Dependency structure

Input Effect sizes and reconciled species names.

Action Count links among effect sizes, references, systems, and species.

Output Dependency tables used to define grouping terms.

Next Check the phylogenetic matrix.

source(here::here("Scripts", "00_packages.R"))
source(here::here("Scripts", "01_paths.R"))
source(here::here("Scripts", "05_phylogeny.R"))

6.1 Input: build the dependency dataset

dat_es <- readRDS(here::here("Rdata", "effect_sizes", "proceed_lnm_safe.rds"))
name_map <- readRDS(here::here("Rdata", "phylogeny", "proceed_name_map.rds"))
dat_mapped <- apply_phylo_name_map(dat_es, name_map)

dat_re_diag <- dat_mapped |>
  dplyr::filter(
    !is.na(sp_ncbi_canonical),
    nzchar(as.character(sp_ncbi_canonical))
  ) |>
  dplyr::transmute(
    es_id_db = factor(es_id_db),
    ref_id   = factor(ref_id),
    sys_id   = factor(sys_id),
    sp_ncbi  = factor(sp_ncbi_canonical)
  ) |>
  tidyr::drop_na() |>
  droplevels()

analysis_counts <- tibble::tibble(
  Unit = c(
    "Contrasts",
    "Effect-size identifiers",
    "References",
    "Study systems",
    "Canonical species"
  ),
  N = c(
    nrow(dat_re_diag),
    dplyr::n_distinct(dat_re_diag$es_id_db),
    dplyr::n_distinct(dat_re_diag$ref_id),
    dplyr::n_distinct(dat_re_diag$sys_id),
    dplyr::n_distinct(dat_re_diag$sp_ncbi)
  )
)

knitr::kable(analysis_counts, caption = "Counts in the dependency dataset.")

Counts in the dependency dataset.

Unit	N
Contrasts	7186
Effect-size identifiers	7186
References	254
Study systems	1533
Canonical species	253

“Canonical species” means unique species labels after taxonomic reconciliation. The current data contain 261 resolved input names but 253 canonical species because eight additional subspecies or duplicate labels collapse into five shared species labels. The next chapter documents this reconciliation and distinguishes the 253 current species from the 265 tips stored in the cached phylogenetic matrix.

6.2 Action: count pairwise links

summarise_links <- function(data, factor_a, factor_b) {
  links <- data |>
    dplyr::distinct(.data[[factor_a]], .data[[factor_b]]) |>
    dplyr::count(.data[[factor_a]], name = "n_b")

  tibble::tibble(
    `Group A` = factor_a,
    `Linked group B` = factor_b,
    `Levels of A` = dplyr::n_distinct(data[[factor_a]]),
    `Levels of B` = dplyr::n_distinct(data[[factor_b]]),
    `A levels linked to one B` = sum(links$n_b == 1),
    `A levels linked to one B (%)` = round(100 * mean(links$n_b == 1), 1),
    `Median B levels per A` = stats::median(links$n_b),
    `Maximum B levels per A` = max(links$n_b)
  )
}

group_vars <- c("es_id_db", "ref_id", "sys_id", "sp_ncbi")

dependency_table <- purrr::map_dfr(group_vars, function(a) {
  purrr::map_dfr(setdiff(group_vars, a), function(b) {
    summarise_links(dat_re_diag, a, b)
  })
})

knitr::kable(
  dependency_table,
  digits = 1,
  caption = "Counts and proportions for each directed pair of grouping variables."
)

Counts and proportions for each directed pair of grouping variables.

Group A	Linked group B	Levels of A	Levels of B	A levels linked to one B	A levels linked to one B (%)	Median B levels per A	Maximum B levels per A
es_id_db	ref_id	7186	254	7186	100.0	1	1
es_id_db	sys_id	7186	1533	7186	100.0	1	1
es_id_db	sp_ncbi	7186	253	7186	100.0	1	1
ref_id	es_id_db	254	7186	23	9.1	8	1284
ref_id	sys_id	254	1533	215	84.6	1	1161
ref_id	sp_ncbi	254	253	235	92.5	1	64
sys_id	es_id_db	1533	7186	1176	76.7	1	1156
sys_id	ref_id	1533	254	1494	97.5	1	10
sys_id	sp_ncbi	1533	253	1533	100.0	1	1
sp_ncbi	es_id_db	253	7186	68	26.9	5	1192
sp_ncbi	ref_id	253	254	200	79.1	1	13
sp_ncbi	sys_id	253	1533	198	78.3	1	379

6.3 Study-system relationships

system_table <- dplyr::bind_rows(
  summarise_links(dat_re_diag, "sp_ncbi", "sys_id"),
  summarise_links(dat_re_diag, "sys_id", "sp_ncbi"),
  summarise_links(dat_re_diag, "ref_id", "sys_id"),
  summarise_links(dat_re_diag, "sys_id", "ref_id")
)

knitr::kable(
  system_table,
  digits = 1,
  caption = "Counts and proportions linking study systems with species and references."
)

Counts and proportions linking study systems with species and references.

Group A	Linked group B	Levels of A	Levels of B	A levels linked to one B	A levels linked to one B (%)	Median B levels per A	Maximum B levels per A
sp_ncbi	sys_id	253	1533	198	78.3	1	379
sys_id	sp_ncbi	1533	253	1533	100.0	1	1
ref_id	sys_id	254	1533	215	84.6	1	1161
sys_id	ref_id	1533	254	1494	97.5	1	10