Skip to contents

Auto-Tune LD Block Detection Parameters

Source: R/run_and_tune.R
tune_LD_params.Rd

Performs a grid search over Big_LD() parameters and selects the combination that minimises, in order of priority:

  1. Unassigned GWAS markers (markers not falling in any block).

  2. Forced assignments (nearest-block fall-back).

  3. Number of blocks (parsimony).

  4. Deviation from target_bp_band (biological plausibility).

If prefer_perfect = TRUE (default), combinations achieving zero unassigned and zero forced assignments are prioritised among the above.

After selecting the best parameter set, tune_LD_params runs run_Big_LD_all_chr on all chromosomes and assigns every GWAS marker to a block, returning the final blocks and assignments.

Usage

tune_LD_params(
  geno_matrix,
  snp_info,
  gwas_df,
  grid = NULL,
  chromosomes = NULL,
  target_bp_band = c(50000, 5e+05),
  parallel = FALSE,
  seed = NULL,
  prefer_perfect = TRUE,
  return_all_perfect = TRUE
)

Arguments

geno_matrix

Numeric matrix (individuals x SNPs; 0/1/2), genome-wide.

snp_info

Data frame with columns SNP, CHR, POS.

gwas_df

Data frame with columns Marker, CHR, POS.

grid

Optional data frame of parameter combinations. Each row is one combination; columns must match parameter names of Big_LD. If NULL (default), a sensible grid over CLQcut (4 values) and min_freq (2 values) is used, giving 8 combinations. Both are treated as hyperparameters following Weber SE et al. (2023, Front. Plant Sci. 14:1217589, doi:10.3389/fpls.2023.1217589 ), who show that no single threshold is universally optimal across datasets and traits.

chromosomes

Optional character vector of chromosome names to include in tuning. NULL uses all chromosomes in snp_info.

target_bp_band

Length-2 numeric vector: preferred median block size range in base pairs. Default c(5e4, 5e5) (50 kb - 500 kb).

parallel

Logical. If TRUE, uses future.apply::future_lapply for parallelism (user must set a future plan before calling). Default FALSE.

seed

Integer seed for reproducibility. Default NULL.

prefer_perfect

Logical. Give priority to parameter sets with zero unassigned / zero forced GWAS markers. Default TRUE.

return_all_perfect

Logical. Include a table of all zero-zero combinations in the returned list. Default TRUE.

Value

A named list:

best_params

Named list of the selected parameters.

score_table

Data frame of all grid combinations and their scores.

perfect_table

Data frame of all zero-zero combinations (or NULL if none found / return_all_perfect = FALSE).

final_blocks

Block table produced with best_params on all chromosomes (output of run_Big_LD_all_chr).

gwas_assigned

Input gwas_df with an added column LD_block. Entries ending in * denote forced assignments.

References

Weber SE, Frisch M, Snowdon RJ, Voss-Fels KP (2023). Haplotype blocks for genomic prediction: a comparative evaluation in multiple crop datasets. Frontiers in Plant Science 14:1217589. doi:10.3389/fpls.2023.1217589

Difabachew YF et al. (2023). Genomic prediction with haplotype blocks in wheat. Frontiers in Plant Science 14:1168547. doi:10.3389/fpls.2023.1168547

See also

run_Big_LD_all_chr, Big_LD()