Finlay-Wilkinson Stability Analysis of Haplotype Effects Across Environments

Estimates the stability of each haplotype block's effect across multiple environments using Finlay-Wilkinson (1963) regression. For each block, the per-environment GEBV contribution is regressed on the environment mean (the environmental index). A regression slope b_i = 1 indicates average stability; b_i > 1 = above-average response (exploits good environments); b_i < 1 = below-average response (robust across environments).

Usage

run_haplotype_stability(
  geno_matrix,
  snp_info,
  blocks,
  blues_list,
  top_n = NULL,
  min_freq = 0.05,
  min_snps = 3L,
  verbose = TRUE
)

Arguments

geno_matrix

Numeric matrix (individuals x SNPs).

snp_info

Data frame with SNP, CHR, POS.

blocks

LD block table from run_Big_LD_all_chr.

blues_list

Named list of named numeric vectors, one per environment: list(env1 = c(id1=val,...), env2 = ...).

top_n

Integer or NULL. Max haplotype alleles per block. Default NULL.

min_freq

Numeric. Minimum haplotype allele frequency. Default 0.05.

min_snps

Integer. Minimum SNPs per block. Default 3L.

verbose

Logical. Default TRUE.

Value

Data frame with one row per block, sorted by CHR and start_bp.

block_id, CHR, start_bp, end_bp

Block coordinates.

b

Finlay-Wilkinson slope. b=1: average stability; b>1: exploits good environments; b<1: robust across environments.

b_se

Standard error of b.

r2_fw

R-squared of the FW regression.

s2d

Deviation mean square (non-linear instability).

stable

Logical; TRUE when b is not significantly different from 1 at alpha=0.05.

References

Finlay KW, Wilkinson GN (1963). The analysis of adaptation in a plant-breeding programme. Australian Journal of Agricultural Research 14(6):742-754.

See also

run_haplotype_prediction, compute_local_gebv

Examples

# \donttest{
data(ldx_geno, ldx_snp_info, ldx_blocks, ldx_blues, package = "LDxBlocks")
# Simulate two environments by splitting the BLUEs
b1 <- setNames(ldx_blues$YLD + rnorm(nrow(ldx_blues), 0, 0.1),
               ldx_blues$id)
b2 <- setNames(ldx_blues$YLD + rnorm(nrow(ldx_blues), 0.5, 0.15),
               ldx_blues$id)
stab <- run_haplotype_stability(
  geno_matrix = ldx_geno,
  snp_info    = ldx_snp_info,
  blocks      = ldx_blocks,
  blues_list  = list(env1 = b1, env2 = b2),
  verbose     = FALSE
)
head(stab[order(stab$b), ])
#>                block_id CHR start.bp end.bp       b b_se r2_fw s2d p_b1 stable
#> 7    block_3_1000_19068   3     1000  19068 -5.3651  NaN     1  NA  NaN  FALSE
#> 6  block_2_86236_105290   2    86236 105290 -3.7255  NaN     1  NA  NaN  FALSE
#> 9   block_3_74532_93854   3    74532  93854 -2.9515  NaN     1  NA  NaN  FALSE
#> 2 block_1_155368_179371   1   155368 179371 -0.1938  NaN     1  NA  NaN  FALSE
#> 5 block_2_161515_180473   2   161515 180473  1.6044  NaN     1  NA  NaN  FALSE
#> 8 block_3_149647_168376   3   149647 168376  1.6153  NaN     1  NA  NaN  FALSE
# }