Compute Local Haplotype GEBV per Block (Tong et al. 2025)

For each LD block, computes the local GEBV (haplotype effect) for every individual by summing the per-SNP additive effects of the alleles they carry within the block:

$$\text{local GEBV}_f = \sum_{t \in f} \left[ h_t \alpha_{1t} + (1-h_t)\alpha_{0t} \right]$$

where $h_t \in \{0, 0.5, 1\}$ is the allele dosage at SNP $t$ scaled to [0,1], $\alpha_{1t}$ is the ALT allele effect, and $\alpha_{0t} = -\alpha_{1t}$ is the REF allele effect (by symmetry of the additive model).

Blocks are ranked by Var(local GEBV) – blocks with high variance contribute strongly to trait differences among individuals and likely harbour causal loci (Tong et al. 2025).

Usage

compute_local_gebv(geno_matrix, snp_info, blocks, snp_effects, scale = TRUE)

Arguments

geno_matrix: Numeric matrix (individuals x SNPs), values 0/1/2/NA.
snp_info: Data frame with columns SNP, CHR, POS.
blocks: Block table from run_Big_LD_all_chr.
snp_effects: Named numeric vector of per-SNP additive effects from backsolve_snp_effects or from a marker model directly.
scale: Logical. If TRUE (default), scale Var(local GEBV) to [0,1] so blocks are comparable across traits and datasets.

Value

A list with two elements:

local_gebv: Numeric matrix (individuals x blocks) of per-block local GEBV values.
block_importance: Data frame with one row per block: block_id, CHR, start_bp, end_bp, n_snps, var_local_gebv, var_scaled, important (logical: scaled variance >= 0.9).

References

Tong J et al. (2025). Haplotype stacking to improve stability of stripe rust resistance in wheat. Theoretical and Applied Genetics 138:267. doi:10.1007/s00122-025-05045-0

Examples

if (FALSE) { # \dontrun{
snp_fx  <- backsolve_snp_effects(my_geno, gebv)
loc     <- compute_local_gebv(my_geno, snp_info, blocks, snp_fx)
# Blocks with scaled variance >= 0.9 are most important
head(loc$block_importance[loc$block_importance$important, ])
} # }