Open eric-czech opened 4 years ago
This isn't supported in Zarr yet but because it is in Dask and the majority of genotyping data sets have major alleles with high frequency, there may be some significant speedups resulting from a sparse in-memory representation.
For reference here's the Zarr spec issue tracking support for sparse layouts: https://github.com/zarr-developers/zarr-specs/issues/48
This isn't supported in Zarr yet but because it is in Dask and the majority of genotyping data sets have major alleles with high frequency, there may be some significant speedups resulting from a sparse in-memory representation.