Where do missing variant/missing variant concordance sites come from?

Hail Query & hailctl
1

What scenario leads to the concordance table entries where neither dataset has a variant? The way the numbers work out, this doesn’t seem to be filtered rows and based on the counts in PCA log messages, I don’t think it’s monomorphic sites.

Also are filtered entries being counted in the concordance table? The documentation isn’t very clear on that point.

2

You’re right – filtered entries are counted as “No Data”. I’ll fix the docs to reflect this. If you want to treat filtered entries as missing GTs, you can do unfilter_entries() before concordance on each side.

3

I didn’t notice unfilter_entries(), but that could be useful. Thanks for that and the clarification!