infection_history_symmetric: Brute force infection history proposal

In seroanalytics/serosolver: Inference Framework for Serological Data

Brute force infection history proposal

Description

Performs a flipping/swapping infection history update step for a matrix of infection histories. 50/50 chance of performing a flip or a swap

Usage

infection_history_symmetric(
  new_inf_hist,
  sampled_indivs,
  age_mask,
  sample_mask,
  proposal_inf_hist_distances,
  n_infs,
  rand_ns,
  proposal_inf_hist_indiv_swap_ratio = 0.5
)

Arguments

new_inf_hist	a matrix of infection histories - rows for individuals, columns for infection epochs. Contents should be 1s and 0s
sampled_indivs	a vector of indices describing rows in the infection history matrix that should be updated
age_mask	a vector (one value for each individual) giving the first infection epoch that an individual could have been exposed in. That is, if an individual was born in the 7th epoch, their entry in age_mask would be 7.
sample_mask	a vector (one value for each individual) giving the last infection epoch that an individual could have been exposed in.
proposal_inf_hist_distances	when performing a move step, how far should two epochs be swapped?
n_infs	number of infection epochs to flip
rand_ns	pre-computed random numbers (0-1) for each individual, deciding whether to do a flip or swap
proposal_inf_hist_indiv_swap_ratio	threshold for deciding if swap or add/remove step

Value

a matrix of infection histories matching the input new_inf_hist

See Also

Other proposals: inf_hist_swap_phi(), inf_hist_swap(), mvr_proposal(), univ_proposal()

Examples

data(example_inf_hist)
data(example_antibody_data)
data(example_antigenic_map)

times <- example_antigenic_map$inf_times
ages <- unique(example_antibody_data[,c("birth","individual")])

## Create age and strain mask
age_mask <- create_age_mask(ages$birth, times)
sample_mask <- create_sample_mask(example_antibody_data, times)

## Index of individuals to be resampled
indivs <- 1:nrow(example_inf_hist)
n_indiv <- length(indivs)

## Parameters controlling proposal sizes for each individual
proposal_inf_hist_distances <- rep(3, n_indiv)
n_infs <- rep(10, n_indiv)

## Pre-compute random numbers
rand_ns <- runif(n_indiv)

new_inf_hist <- infection_history_symmetric(example_inf_hist, indivs,age_mask ,sample_mask, proposal_inf_hist_distances, n_infs, rand_ns, 0.5)

seroanalytics/serosolver documentation built on Dec. 14, 2024, 5:33 a.m.