v2.1.4

Fix bug preventing to go to sorted() structure within Amatrix for some pedigree files

v2.1.3

Add ASV option to Amatrix, Gmatrix, Hmatrix
Change citation
Reverse back default arguments to match with historical default (thanks ASRgenomics check)

v2.1.2

Fix issue with pedigree ordering, fix Mcheck minor

v2.1.0

• snp.check now is Mcheck
• reviewed tutorial/functions

v2.0.4

Major updates:

• Revisited and updated tutorial (https://cran.r-project.org/web/packages/AGHmatrix/vignettes/Tutorial_AGHmatrix.html)
• Earlier autopolyploid Gmatrix() (for VanRaden method) was being scaled by the sampling variance, now the default is to scale by the parametric format (i.e.: ploidypq) which is the most reported scaling way in the literature. Both give same output under infinite number of SNPs. As it is just a scaling factor, the prediction accuracies out of this revisited GRM are the same.
• Gmatrix() imputation. The default now is to input missing value by the mean of each marker. Earlier was imputing the missing data by a unique global mean across all markers (not ideal approach that was completely overlooked by me). There is an option to input by median too (by marker or global).
• New Gmatrix() option to include SNP weights. You can indicate a vector with weights for each marker (from a GWAS output for example).
• New filterpedigree() function. Given a pedigree and a vector of individuals, it reduces the pedigree to keep just the vector of individuals and their ancestral entries. This speeds up and reduces the required specs to build the A matrix while keeping the same output. Some people from animal breeding were facing problems to build the A matrix with really large pedigrees (>100K entries) while the interest was just on the relationship of just some hundred individuals, this approach has been solving the problem.
• New AmatrixPolyCross() function. It is an expansion of the Amatrix() function that allows to have possible parents out of a pool of parents. Example, offspring of [mom A x (dad B or dad C)] will have 0.5 of relatedness with mom A and 0.25 of relatedness with dad B and with dad C; and this probability is propagated in the recurrent algorithm in the pedigree. It allows fix mother x pool of dad, or pool of parents equally possible.
• New Amatrix()'s benchmark covering [RAM x pedigree size x computational time] located at the tutorial's end.

Let me know if you have any question or suggestion.