In benmoran11/rubias: Bayesian Inference from the Conditional Genetic Stock Identification Model

Have just developed a way to do haploid markers and I just want to do a quick test to make sure that it is working the way I would hope it might.

Make some data that are quite simple:

simp <- tibble(sample_type = "reference", 
       repunit = "Boing", 
       collection = paste0("coll", rep(1:3, each = 10)), 
       indiv = paste0("indiv", rep(1:3, each = 10), "_", 1:10),
       hap1 = as.integer(c(rep(1,10), rep(1,5), rep(2,5), rep(3,10))),
       hap1.1 = NA
       )

This is pretty straightforward. Coll 1 and 3 are fixed for types 1 and 3 respecively. Coll 2 has 5 1's and 5 2's.
Each pop has ten indivs in it.

Let's use a scaled constant prior which will give us 1/3, 1/3, 1/3.

So, any individual in collection 1 to itself should have a logl of 9.33333 / 10 = 0.93. From collection 1 to 2 should be the same for each individual...etc.

From Collection 2 with a 1 allele to collection 1 should be 10.3/11

check_it <- self_assign(simp, 5) %>%
  mutate(like = exp(log_likelihood))

And, when you check all those out, they look solid.

Cool, now, let's do the same, but add in a bunch of loci with random alleles. I basically want to check that we get the same result no matter how these loci are ordered in the data set.

set.seed(55)
loc3a <- simp %>%
  mutate(hap0 = sample(1:5, 30, replace = TRUE),
         hap0.1 = sample(1:5, 30, replace = TRUE),
         hap2 = sample(1:5, 30, replace = TRUE),
         hap2.1 = sample(1:5, 30, replace = TRUE))

Then do it up:

a <- self_assign(loc3a, 5)

loc3b <- loc3a[, c(1, 2, 3, 4, 7, 8, 5, 6, 9, 10)]
b <- self_assign(loc3b, 5)

loc3c <- loc3a[, c(1, 2, 3, 4, 9, 10, 7, 8, 5, 6)]
c <- self_assign(loc3c, 5)

identical(a, b)
identical(a, c)

Now, make some missing data

loc3m <- loc3b
loc3m[1, 7] <- NA
loc3m[2, c(5, 6)] <- NA
loc3m[3, c(5,6, 7)] <- NA
m3 <- self_assign(loc3m, 5)

benmoran11/rubias documentation built on Feb. 1, 2024, 10:52 p.m.