Open tsalo opened 3 weeks ago
one possible gotcha here is that a very large number of these matrices get created. Each atlas gets a matrix for each weighting type (eg sift, average scalar, etc) and connectivity calculation method (pass, end, etc).
It's a hard one. In XCP-D there are a bunch of correlation matrices that we just put in separate files. We end up with a lot of files, but it's still manageable.
Summary
BEP017: Relationship matrices defines how correlation matrices should be defined. We can still compress the
relmat
files if we adopt BEP017, whether withtsv.gz
, zarr, h5, or parquet.