predict.DNAmixture {DNAmixturesLite}

Various probabilities in a fitted DNA mixture model

Description

IMPORTANT: This is the DNAmixturesLite package, which is intended as a service to enable users to try DNAmixtures without purchasing a commercial licence for Hugin. When at all possible, we strongly recommend the use of DNAmixtures rather than this lite-version. See https://dnamixtures.r-forge.r-project.org/ for details on both packages.

While the lite-version seeks to provide the full functionality of DNAmixtures, note that computations are much less efficient and that there are some differences in available functionality. Be aware that the present documentation is copied from DNAmixtures and thus may not accurately describe the implementation of this lite-version.

Usage

## S3 method for class 'DNAmixture'
predict(
  object,
  pars,
  dist = c("joint", "conditional", "prequential"),
  markers = object$markers,
  by.allele = TRUE,
  initialize = TRUE,
  ...
)

Arguments

object

A DNAmixture object

pars

Array of model parameters

dist

One of "joint", "conditional", and "prequential". If there are only known contributors, these are all the same since, under the model, peak heights are condtionally independent given profiles of the contributors.

markers

The set of markers of interest

by.allele

If dist = "prequential" then the order in which we condition on mixtures and alleles matters. by.allele = TRUE will proceed through alleles in increasing repeat number, and for each allele condition on one mixture at the time. If FALSE, the conditioning is done by mixtures and then alleles within these.

initialize

By default predict removes all entered evidence from the networks in object. Setting initialize = FALSE should be done with care, and it is up to the user to ensure that the returned probabilities are meaningful.

...

Not used

Details

For a mixture with unknown contributors, the probabilities are computed with respect to one of three distributions. Let height be the matrix of peak heights with columns height1, ..., heightR. For a peak at allele a in the mixture r, the three choices of distributions are

"joint"

Default. No conditioning on observed peak heights.

"conditional"

Conditional on height[-a, -r], i.e. on heights for all peaks, except the one under consideration.

"prequential"

Conditional on height[1:(a-1), 1:(r-1)], i.e. on heights for all peaks "before" the peak under consideration (see argument by.allele for details).

If all contributors are known, the three distributions are the same due to independence of the peak heights.

Value

A list with one data.frame per marker containing various probabilities for diagnostics

unseen

The probability of not seeing a peak, i.e. no peak or a peak falling below the threshold

seen

The probability of seeing the allele

smaller

The probability of seeing a smaller peak than the one observed

larger

The probability of seeing a larger peak than the one observed

Author(s)

Therese Graversen

Examples

data(MC15, MC18, USCaucasian)
mix <- DNAmixture(list(MC15, MC18), C = list(50,50), k = 3, K = c("K1", "K3", "K2"),
database = USCaucasian)
p <- mixpar(rho = list(30, 30), eta = list(30, 30), xi = list(0.08,0.08),
            phi = list(c(K2 = 0.1, K3 = 0.2, K1 = 0.7), c(K2 = 0.1, K3 = 0.2, K1 = 0.7)))
pred <- predict(mix, p)
pred$VWA