rubias (version 0.3.4)

avg_coll2correctRU: Get the average within-RU assignment rate for each collection

Description

This function takes a matrix of scaled genotype likelihoods for a group of individuals of known origin, and calculates the average rate at which individuals in a particular collection are assigned to the correct reporting unit.

Usage

avg_coll2correctRU(SL, coll, RU_starts, RU_vec)

Value

avg_coll2correctRU returns a vector of length nrow(SL), where each element represents the average proportion of fish from the corresponding collection which are correctly assigned to the proper collection, or misassigned to another collection within the same reporting unit. This is distinct from the rate of correct assignment at the collection level, which is too low and variable to serve as a stable metric for omega scaling.

Arguments

SL

a scaled likelihood matrix; each column should sum to one, and represent the probability of assignments to each collection (row) for a particular individual

coll

a vector of the collection of origin indices of the individuals (length = ncol(SL))

RU_starts

a vector delineating starting indices of different reporting units in RU_vec

RU_vec

a vector of collection indices, organized by reporting unit

Details

The average rate of correct within-reporting unit assignment is proportional to reporting-unit-level bias in the posterior probability for this collection; if the correct assignment rate is high relative to other collections, it will be upwardly biased, and vice versa. The inverse of this vector is used to scale Dirichlet draws of omega during misassignment-scaled MCMC.

Examples

locnames <- names(alewife)[-(1:16)][c(TRUE, FALSE)]
ploidies <- rep(2, length(locnames))
names(ploidies) <- locnames

params <- tcf2param_list(alewife, 17, ploidies = ploidies)
SL <- geno_logL(params) %>% exp() %>% apply(2, function(x) x/sum(x))
correct <- avg_coll2correctRU(SL, params$coll, params$RU_starts, params$RU_vec)