Create and return TSENAT configuration — TSENAT

Builds a configuration list for use with TSENAT(). Allows specifying analysis parameters once and reusing across multiple analyses.

Usage

TSENAT_config(
  q = 1,
  condition_col = "condition",
  subject_col = NULL,
  sample_col = "sample",
  paired = FALSE,
  control = NULL,
  p_threshold = 0.05,
  fdr_threshold = 0.05,
  significance_threshold = 0.05,
  bootstrap = FALSE,
  nboot = 1000,
  bootstrap_method = "percentile",
  stringency = "medium",
  nthreads = 1,
  norm = TRUE,
  bootstrap_ci = 0.95,
  bootstrap_include_diagnostics = TRUE,
  min_valid_frac = 0.75,
  norm_method = NULL,
  pseudocount = 0,
  shrinkage = "none",
  lm_method = "gam",
  lm_pcorr = "BH",
  jis_use_lm_fdr = TRUE,
  divergence_ci = 0.95,
  assumptions_checks = "all",
  ...
)

Arguments

q: numeric. Q-value(s) for Tsallis entropy (single value or vector). Default: 1.0 (Shannon entropy). Usage: calculate_diversity/divergence use this for spectrum computation (if vector) or as default fallback (if single).
condition_col: character. Column name in colData containing conditions. Default: 'condition'.
subject_col: character. Column name in colData containing subject IDs (for paired designs). Default: NULL.
sample_col: character. Column name in colData containing sample IDs. Default: 'sample'.
paired: logical. Whether samples are paired/repeated measures. Default: FALSE.
control: character. Reference/control group label. Default: NULL.
p_threshold: numeric. Raw p-value threshold. Default: 0.05.
fdr_threshold: numeric. FDR-adjusted p-value threshold. Default: 0.05.
significance_threshold: numeric. Significance cutoff. Default: 0.05.
bootstrap: logical. Enable bootstrap CIs. Default: FALSE.
nboot: integer. Bootstrap resamples for CIs. Default: 1000.
bootstrap_method: character. Bootstrap method: 'percentile' or 'bca'. Default: 'percentile'.
stringency: character. Filtering stringency: 'lenient', 'medium', 'severe'. Default: 'medium'.
nthreads: integer. Parallel threads. Default: 1.
norm: logical. Enable normalization. Default: TRUE.
bootstrap_ci: numeric. Confidence level (0-1). Default: 0.95.
bootstrap_include_diagnostics: logical. Include diagnostics. Default: TRUE.
min_valid_frac: numeric. Min valid replicate fraction. Default: 0.75.
norm_method: character. Normalization: NULL, 'zscore', 'log_odds_ratio', 'relative_reference'. Default: NULL.
pseudocount: numeric. Pseudocount for sparse data. Default: 0.
shrinkage: character. Variance reduction: 'none' or 'empirical_bayes'. Default: 'none'.
lm_method: character. LM method: 'gam', 'lmm', 'fpca', 'gee'. Default: 'gam'.
lm_pcorr: character. P-value correction: 'BH', 'bonferroni', 'hochberg', 'holm'. Default: 'BH'.
jis_use_lm_fdr: logical. Filter jackknife genes using LM p-values. Default: TRUE.
divergence_ci: numeric. Confidence level for divergence CIs. Default: 0.95.
assumptions_checks: character. Which assumptions to test (default: 'all'). Presets: - 'rank': core assumption checks (exchangeability, monotonicity, consistency) - 'all': all checks including method-specific diagnostics (GAM, GEE, LMM, FPCA) Explicit: character vector like c('exchangeability', 'monotonicity').
...: Additional configuration parameters (stored as-is).

Value

list with class TSENATConfig containing all specified parameters.

Details

Configuration is stored in the TSENATAnalysis@config slot and used by wrapper functions to configure analysis behavior. Note: Statistical tests (Wilcoxon, shuffle) work on a single q-value, so only one q-value is specified in config.

Examples

# Default config with standard parameters (point estimates only)
cfg <- TSENAT_config()

# For Wilcoxon/shuffle tests (single q-value required in config)
cfg <- TSENAT_config(
  q = 1.0,                          # Shannon entropy - for rank tests
  condition_col = 'treatment',
  control = 'untreated'
)

# For Scheirer-Ray-Hare rank tests (multiple q-values)
cfg <- TSENAT_config(
  q = seq(0, 2, by = 0.5),          # Multiple q-values for spectrum or advanced testing
  condition_col = 'treatment',
  control = 'untreated'
)

# With bootstrap CIs for uncertainty quantification (recommended)
cfg <- TSENAT_config(
  bootstrap = TRUE,                # Enable bootstrap confidence intervals
  bootstrap_method = 'bca',         # Bias-corrected (better for skewed entropy)
  nboot = 1000,                     # 1000 resamples
  bootstrap_ci = 0.95               # 95% CI
)

# Custom with paired analysis, strict filtering, and normalization
cfg <- TSENAT_config(
  q = 1.0,                          # Shannon entropy
  condition_col = 'treatment',
  subject_col = 'subject_id',
  paired = TRUE,
  control = 'untreated',
  stringency = 'severe',            # High-confidence transcripts only
  norm_method = 'zscore',           # Cross-study standardization
  shrinkage = 'none',               # Empirical estimates
  bootstrap = TRUE,
  bootstrap_method = 'bca',
  nboot = 5000,                     # Higher precision
  pseudocount = 0,                  # Disabled by default; set > 0 to add pseudocount
  significance_threshold = 0.01     # Stricter significance level
)