Pairwise Correlation Analysis for Thermal and External Traits — correlate_thermal

Computes pairwise correlations between a set of thermal variables and a set of external variables. It handles missing data gracefully, performs multiple testing corrections, and returns detailed statistical summaries along with correlation and p-value matrices.

Usage

correlate_thermal_traits(
  data,
  thermal_vars,
  external_vars,
  method = "spearman",
  adjust_method = "BH",
  use = "complete.obs"
)

Arguments

data: A `data.frame` containing the variables to be analyzed.
thermal_vars: A character vector specifying the column names of the thermal variables in `data`.
external_vars: A character vector specifying the column names of the external variables in `data`.
method: A character string indicating which correlation coefficient to compute. Can be either `"spearman"` (default) or `"pearson"`.
adjust_method: A character string specifying the method for multiple testing correction. Options are `"holm"`, `"hochberg"`, `"hommel"`, `"bonferroni"`, `"BH"` (default), `"BY"`, `"fdr"`, or `"none"`.
use: A character string indicating how to handle missing values. Can be `"complete.obs"` (default, uses only rows with complete data across all selected variables) or `"pairwise.complete.obs"` (uses complete cases on a pair-by-pair basis).

Value

A list containing the following elements:

results: A data frame with detailed pairwise statistics including variable names, sample size (`n`), correlation coefficient (`cor`), raw p-value (`p`), adjusted p-value (`p_adj`), direction, significance label, and any warnings.
cor_matrix: A numeric matrix of the calculated correlation coefficients.
p_matrix: A numeric matrix of the raw p-values.
padj_matrix: A numeric matrix of the adjusted p-values.
method: The correlation method used.
adjust_method: The p-value adjustment method used.
use: The missing data handling method used.
use_note: A descriptive note regarding how missing data was actually handled.

Examples

set.seed(1234)
df <- data.frame(
  Max = rnorm(50, 35, 2),
  Min = rnorm(50, 15, 3),
  Weight = rnorm(50, 20, 5),
  Glu = runif(50, 5, 8)
)

# Introduce a few NA values to test missing data handling
df$Weight[c(2, 10)] <- NA

# Run the correlation analysis
res <- correlate_thermal_traits(
       data = df,
       thermal_vars = c("Max", "Min"),
       external_vars = c("Weight", "Glu"),
       method = "spearman",
       adjust_method = "BH",
       use = "pairwise.complete.obs"
       )
# View the detailed results data frame
head(res$results)
#>   thermal_var external_var   method  n         cor          p     p_adj
#> 1         Max       Weight spearman 48  0.10790274 0.46538913 0.5197083
#> 2         Max          Glu spearman 50  0.27270108 0.05536452 0.2214581
#> 3         Min       Weight spearman 48  0.16261398 0.26946508 0.5197083
#> 4         Min          Glu spearman 50 -0.09320528 0.51970832 0.5197083
#>   direction significance warning
#> 1  positive           ns    <NA>
#> 2  positive           ns    <NA>
#> 3  positive           ns    <NA>
#> 4  negative           ns    <NA>

# View the correlation matrix
print(res$cor_matrix)
#>        Weight         Glu
#> Max 0.1079027  0.27270108
#> Min 0.1626140 -0.09320528