The likelihood of a relationship graph (RG) can be decomposed over markers, as we assume that marker data is conditionally independent across markers given the RG. The likelihood of a RG for one marker is then found by integrating over all possible IBD (identity-by-descent) partitions that are consistent with the RG. The probability distribution of these IBD partitions are always uniform. Since different RGs may use the same IBD, the likelihood of IPs for each marker are stored in hash tables for improved computational efficiency.
Arguments
- MOIs
A numeric vector specifying, for each infection, the number of distinct parasite genotypes, a.k.a. the multiplicity of infection (MOI).
- fs
List of allele frequencies as vectors. Names of the list must match with the marker names in
alleles_per_m.- alleles_per_m
List of allele assignments as dataframes for each marker. Each column corresponds to a genotype and each row corresponds to an allele assignment.
- progress.bar
Boolean for printing progress bars.
Value
List of all relationship graph objects, including their log-likelihoods as a variable under each object.
Examples
# 1 marker, 2 infections, MOIs = 2, 1
MOIs <- c(2, 1)
fs <- list(
m1 = stats::setNames(c(0.4, 0.6), c("A", "B")),
m2 = stats::setNames(c(0.2, 0.8), c("C", "D"))
)
al_df1 <- as.data.frame(matrix(c("A", "B", "B"), nrow = 1))
al_df2 <- as.data.frame(matrix(c(
"C", "D", "C",
"D", "C", "C"
), nrow = 2, byrow = TRUE))
colnames(al_df1) <- colnames(al_df2) <- paste0("g", 1:3)
alleles_per_m <- list(m1 = al_df1, m2 = al_df2)
RGs <- RG_inference(MOIs, fs, alleles_per_m)
#> Number of valid relationship graphs (RGs) is 9
#> ========
#> =========
#> ========
#> =========
#> ========
#> =========
#> ========
#> =========
#> =========
#> |
#> Computing log p(Y|RG) for 9 RGs
#> ========
#> =========
#> ========
#> =========
#> ========
#> =========
#> ========
#> =========
#> =========
#> |