sraStartingvalues

These functions aim at providing educated starting values for the parameters to estimate through sra models.

 models

 internal

Artificial selection through selective breeding is an efficient way to induce changes in traits of interest in experimental populations. This package (sra) provides a set of tools to analyse artificial-selection response datasets. The data typically feature for several generations the average value of a trait in a population, the variance of the trait, the population size and the average value of the parents that were chosen to breed. Sra implements two families of models aiming at describing the dynamics of the genetic architecture of the trait during the selection response. The first family relies on purely descriptive (phenomenological) models, based on an autoregressive framework. The second family provides different mechanistic models, accounting e.g. for inbreeding, mutations, genetic and environmental canalization, or epistasis. The parameters underlying the dynamics of the time series are estimated by maximum likelihood. The sra package thus provides (i) a wrapper for the R functions mle() and optim() aiming at fitting in a convenient way a predetermined set of models, and (ii) some functions to plot and analyze the output of the models.

Arnaud Rouzic

Selection Response Analysis

sraStartingvalues function

<dl> <dt>parameter</dt>
<dd>A character string corresponding to the name of a parameter to consider, e.g. <code>"mu0"</code> or <code>"logepsilon"</code>. Depending on the parameter, starting values can be estimated from the dataset, or a reasonable arbitrary value may be returned.</dd> <dt>sradata</dt>
<dd>A data object generated by <code>sraData</code>.</dd> <dt>rand</dt>
<dd>Amount of randomness to add to the parameter value. Setting a value different from 0 leads to a different answer for each function call, which can be useful when trying different vectors of staring values to solve problematic convergence cases. In general, values &gt; 0.2 provide starting values that are too extreme to be useful.</dd></dl>

Arguments

Provides reasonable starting values for sra model fit (internal function) — sraStartingvalues

<dl>

 <dt>parameter</dt>
<dd>A character string corresponding to the name of a parameter to consider, e.g. <code>"mu0"</code> or <code>"logepsilon"</code>. Depending on the parameter, starting values can be estimated from the dataset, or a reasonable arbitrary value may be returned.</dd>

 <dt>sradata</dt>
<dd>A data object generated by <code>sraData</code>.</dd>

 <dt>rand</dt>
<dd>Amount of randomness to add to the parameter value. Setting a value different from 0 leads to a different answer for each function call, which can be useful when trying different vectors of staring values to solve problematic convergence cases. In general, values &gt; 0.2 provide starting values that are too extreme to be useful.</dd>

</dl>

sraStartingvalues: Provides reasonable starting values for sra model fit (internal function)

Description

Usage

Arguments

See Also