Summarize Grid Simulation Results — summarize_grid

Summarize results from run_grid_simulation(), filtering by parameter ranges and computing aggregated performance metrics. Provides detailed summaries at both the setting level and across all selected settings.

Usage

summarize_grid_results(
  grid_results_list,
  n_sequences_range = NULL,
  seq_length_range = NULL,
  min_na_range = NULL,
  max_na_range = NULL,
  level_context = 0.05,
  print_output = TRUE,
  print_aggregated_overall = TRUE,
  print_aggregated_edges = TRUE,
  print_settings_summary = TRUE,
  num_rows_range = NULL,
  max_seq_length_range = NULL
)

analyze_grid_results(...)

Arguments

grid_results_list: List output from run_grid_simulation().
n_sequences_range: Numeric vector of length 2 (min, max) or NULL. Filter settings by n_sequences.
seq_length_range: Numeric vector of length 2 (min, max) or NULL. Filter settings by seq_length.
min_na_range: Numeric vector of length 2 (min, max) or NULL. Filter settings by min_na.
max_na_range: Numeric vector of length 2 (min, max) or NULL. Filter settings by max_na.
level_context: Numeric. Significance level used for calculations (for context in output). Default: 0.05.
print_output: Logical. Master switch for console printing. Default: TRUE.
print_aggregated_overall: Logical. Print averaged overall performance metrics across selected settings. Default: TRUE.
print_aggregated_edges: Logical. Print aggregated edge significance summary. Default: TRUE.
print_settings_summary: Logical. Print summary table for each selected setting. Default: TRUE.
num_rows_range: Deprecated. Use n_sequences_range instead.
max_seq_length_range: Deprecated. Use seq_length_range instead.
...: Arguments passed to summarize_grid_results.

Value

A list containing:

n_selected

Number of settings matching the filter criteria.

aggregated_summary

List with:

overall_performance: Averaged metrics across settings.
edge_significance: Aggregated edge-level statistics.

selected_settings_summary_df

Data frame with per-setting metrics calculated from total TP/TN/FP/FN counts.

compiled_individual_runs

List with detailed run-level data:

all_raw_summaries: Combined bootstrap summaries.
all_per_edge_performance: Combined per-edge results.
run_level_performance_metrics: Metrics per run.
setting_level_summary_stats: Mean/Median/SD of run metrics.

Returns NULL (invisibly) if no settings match.

Details

The function performs comprehensive analysis:

Filtering: Selects settings where all parameters fall within specified ranges.

Aggregation from Input Summaries: Extracts and averages overall performance and edge significance from the original aggregated_summary in each setting.

Run-Level Metrics: Recomputes TP/TN/FP/FN counts from per-edge data, then calculates Sensitivity, Specificity, FPR, FNR, Accuracy, and MCC per run.

Setting-Level Metrics: Two approaches:

Mean/Median/SD of run-level metrics.
Metrics computed from total counts across all runs (more robust).

The selected_settings_summary_df uses approach #2 for the most accurate overall metrics per setting.

Examples

if (FALSE) { # \dontrun{
# After running grid simulation
grid_results <- run_grid_simulation(...)

# Analyze all results
analysis <- summarize_grid_results(grid_results)

# Filter to specific parameter ranges
analysis_filtered <- summarize_grid_results(
  grid_results,
  n_sequences_range = c(100, 200),
  seq_length_range = c(30, 50),
  max_na_range = c(0, 5)
)

# Access the detailed run-level metrics
run_metrics <- analysis$compiled_individual_runs$run_level_performance_metrics

# Access setting-level summary
setting_summary <- analysis$selected_settings_summary_df

# Old parameter names still work
analysis_filtered <- summarize_grid_results(
  grid_results,
  num_rows_range = c(100, 200),
  max_seq_length_range = c(30, 50)
)
} # }