search

AlphaGenes / AlphaPeel

AlphaPeel: calling, phasing, and imputing genotype and sequence data in pedigrees
MIT License
2 stars 11 forks source link

Add accuracy test for other output in addition to allele dosage #80

Open gregorgorjanc opened 1 year ago

gregorgorjanc commented 1 year ago

I believe at the moment we only look at accuracy of allele dosage imputation (in the accuracy tests).

Recent work on functionality tests from @XingerTang shows some unexpected behaviour for calling alleles in haplotypes (phase) (issue incoming), suggesting we should look at accuracy tests for other AlphaPeel outputs, not just the allele dosages.

The amount of output could be quite large since we now look at accuracy per individual and marker and by generation with multiple methods. Adding more outputs will triple/quadruple the amount of outputs! How do we handle this?

XingerTang commented 1 year ago

@gregorgorjanc I don't think the amount of the output would be a big problem. So far, the time required for the calculation of the accuracy is still much less than the time required for the actual peeling, and the different output formats for the peeling would be generated regardless of whether it would be printed out. The problem that concerned me is that we only have true genotypes for the data, so we cannot evaluate the accuracy of other output formats with only true genotypes.

gregorgorjanc commented 1 year ago

@XingerTang we can resimulate the data, which will give us genotypes, haplotypes, and segregation information. In addition, we will save the simulation script so we can always resimulate if needed;)

XingerTang commented 1 year ago

@gregorgorjanc That is great! Then it won't take a lot of time to develop the accuracy tests for other output formats.

gregorgorjanc commented 1 year ago

@XingerTang can you initiate work on an AlphaSimar simulation based on the AlphaSimR MOOC and the current pedigree structure for the accuracy test?

I can help with feedback on your initial R script.

gregorgorjanc commented 1 year ago

@XingerTang can you initiate work on an AlphaSimar simulation based on the AlphaSimR MOOC and the current pedigree structure for the accuracy test?

I can help with feedback on your initial R script.

XingerTang commented 1 year ago

@gregorgorjanc Sure! I can do that.