Changelog

Source: NEWS.md

genomicMateSelectR 0.2.0

Upgrade. Combine predCrossVar with other functions for GS pipeline and add high level wrapping functions that implement/automatic cross predictions, and more.

genomicMateSelectR 0.1.0

See predCrossVar, the OG codebase.

genomicMateSelectR 0.0.0

Development To Do

  • Improve input format for SNP-effects into predictCrosses(). More flexible SNP-effect list-column naming (instead of opinionated e.g. allelesubeffectlist)? Particularly for the DirDom model.

  • Improve phenotype input e.g. for runGenomicPredictions() and runParentWiseCrossVal().

    • Currently demands a number of variables, be named in particular weights, owing to the two-stage genomic prediction process I use for NextGen Cassava. In that Cassava pipeline, I’ve used de-regressed BLUPs as response variable for genomic prediction and weighted error variances according to a weighting scheme devised by Garrick et al. 2009. Users can subvert this, but are currently required to use certain naming: drgBLUP for the response data, BLUP for data to-be-used as validation values in cross-validation, and a column of weights (WT) that could just be set to 1 if users aren’t doing weighted anlaysis.

  • What would make the package BrAPI compliant?

  • SNP marker ID naming conventions are probably too strict; add flexibility / robustness

  • More flexibility for the crosses2predict() function: reciprocal crosses? don’t include selfs?

  • Change predCrossMeans() so that it can accept either dosages or a haploMat. Currently, predictCrosses() requires users to supply both dosages and haploMat . Changing predCrossMeans() in this way will allow users of predictCrosses() to supply only the haploMat if they are predicting both cross means and variances, or only dosages if predicting only means.

  • modelType="AD_geno" or something like that –> genotypic additive-dominance partition allowing prediction of μTGV in addition to μBV but without the genome-wide directional dominance part implemented in modelType="DirDom".

  • When you set getMarkEffs=TRUE in runGenomicPredictions(), tell the user that the function is erroring because they forgot to input the dosage matrix via dosages=.