popImprovByMateSel.RdFunction to improve a simulated breeding population by one cycle.
Also specify selCritPop="genomicMateSelCrit" and use the following new bsp arguments to control the predictions and selection:
modelType: modelTypes: "A", "AD", "DirDom"
crossSelCrit: "MeanBV" (modelTypes: "A", "AD", "DirDom"); "MeanTGV" (only with modelType="DirDom"); "UCparent" (modelTypes: "A", "AD", "DirDom"); "UCvariety" (modelTypes: "AD", "DirDom")
propSel: 0-1, prop. of predicted crosses to select, used ONLY to calc the standardized selection intensity (i) and subsequently the usefulness criteria (\(\mu + i \times \sigma\)).
popImprovByMateSel(records, bsp, SP)
| records | The breeding program |
|---|---|
| bsp | A list of breeding scheme parameters |
| SP | The AlphaSimR SimParam object |
A records object with a new F1 Pop-class object of progeny coming out of a population improvement scheme
My changes:
nTrainPopCycles: draw training pop clones only from this number of recent cycles.
nYrsAsCandidates: candidates for selection only from this number of recent years
maxTrainingPopSize: From the lines in the most recent cycles (indicated by nTrainPopCycles),
subsample this number of lines for training data. This is in addition to the "check" (bsp$checks@id)
and the lines indicates as selection `candidates` according to the setting of `nYrsAsCandidates`.
All "historical" data will always be used, but the number of maximum training lines will be held constant.
Replaces the stage-specific `bsp$trainingPopCycles`, which will be unused in this pipeline, but not deleted from the package.
This function uses phenotypic records coming out of the product pipeline to choose individuals as parents to initiate the next breeding cycle