There are situations where we may want to generate a set of results with specific categories omitted (e.g. because we're running a singleton dataset, and a category that works well for trios may be overwhelmed by noise). We should have a config entry which can remove some categories from consideration.
Plan - do this when generating the variant objects instead of at the Hail stage
The Hail stage should still generate all valid labels, and ignoring them should be handled in the downstream, faster running phase of the process.
There are situations where we may want to generate a set of results with specific categories omitted (e.g. because we're running a singleton dataset, and a category that works well for trios may be overwhelmed by noise). We should have a config entry which can remove some categories from consideration.
Plan - do this when generating the variant objects instead of at the Hail stage
The Hail stage should still generate all valid labels, and ignoring them should be handled in the downstream, faster running phase of the process.