In various places it would be useful to have a quick way of getting well-formed VCZ data, and (maybe) an easy way of comparing two VCZ datasets for equality. The basic functionality would be the same as the sgkit.simulate_genotype_call_dataset, but could be simpler (i.e, using more regular patterns like np.linspace or something to generate the genotypes).
The input parameters to this seems obvious enough, but it's not clear to me what the output should be. Options:
- Dictionary mapping names to arrays
- Zarr group
- Xarray dataset
- Some custom "builder" class, which enables additional modifications.
The dictionary option seems like the simplest and most flexible interchange format without tying to external APIs, and would probably be what I'd lean towards. However, there is a need for some general infrastructure for defining a VCZ dataset etc which is currently lacking, that this should maybe integrate with?
Any thoughts?
In various places it would be useful to have a quick way of getting well-formed VCZ data, and (maybe) an easy way of comparing two VCZ datasets for equality. The basic functionality would be the same as the sgkit.simulate_genotype_call_dataset, but could be simpler (i.e, using more regular patterns like np.linspace or something to generate the genotypes).
The input parameters to this seems obvious enough, but it's not clear to me what the output should be. Options:
The dictionary option seems like the simplest and most flexible interchange format without tying to external APIs, and would probably be what I'd lean towards. However, there is a need for some general infrastructure for defining a VCZ dataset etc which is currently lacking, that this should maybe integrate with?
Any thoughts?