Split phased haplotypes into K population-specific dosage matrices

Generalises split_haplotype to an arbitrary number of ancestries. Each haplotype's allele is routed to the dosage pool whose code matches its local ancestry call. Haplotypes with unrecognised codes contribute 0 to all pools. NA genotypes are treated as 0 (reference allele).

split_haplotype_multi(
  gt_hap0,
  gt_hap1,
  anc_hap0,
  anc_hap1,
  pop_codes = c(AFR = 0L, EUR = 1L)
)

Arguments

gt_hap0

Integer matrix (variants x samples) of haplotype-0 alleles (0/1).

gt_hap1

Integer matrix (variants x samples) of haplotype-1 alleles (0/1).

anc_hap0

Integer matrix (variants x samples) of haplotype-0 ancestry codes.

anc_hap1

Integer matrix (variants x samples) of haplotype-1 ancestry codes.

pop_codes

Named integer vector mapping population label to ancestry code, e.g. c(AFR = 0L, EUR = 1L, NAT = 2L). Codes must match those used in the MSP file (RFMix default: 0-based integers).

Value

Named list of K numeric matrices (variants x samples), one per entry in pop_codes, in the same order. List names equal the names of pop_codes.

Details

For the two-population case split_haplotype_multi and split_haplotype are equivalent; prefer split_haplotype_multi for new code that may later be extended to three or more ancestries.

See also

split_haplotype for the two-population convenience wrapper, split_diploid for unphased (diploid-code) splitting.

Examples

# Three-population panel: AFR=0, EUR=1, NAT=2
gt0 <- matrix(c(1L, 0L, 0L, 1L, 1L, 0L), nrow = 3, ncol = 2)
gt1 <- matrix(c(0L, 1L, 1L, 0L, 0L, 1L), nrow = 3, ncol = 2)
a0  <- matrix(c(0L, 1L, 2L, 0L, 1L, 2L), nrow = 3, ncol = 2)
a1  <- matrix(c(1L, 2L, 0L, 2L, 0L, 1L), nrow = 3, ncol = 2)
out <- split_haplotype_multi(gt0, gt1, a0, a1, c(AFR = 0L, EUR = 1L, NAT = 2L))
names(out)   # "AFR" "EUR" "NAT"
#> [1] "AFR" "EUR" "NAT"
out$NAT      # native-ancestry dosage matrix
#>      [,1] [,2]
#> [1,]    0    0
#> [2,]    1    0
#> [3,]    0    0