Any comment on MAF filtering for XP-nSL/XP-EHH results

[biozzq]

Dear scikit-allel developers,

I am working on calculating XP-nSL and XP-EHH using scikit-allel, and I want to confirm the correct MAF filtering practice, referencing comments from the selscan developer:

1. The selscan author stated: "filtering for MAF when applying the two population statistics (XP-EHH/XP-nSL) hindered power quite a bit. This is why I didn't implement it in selscan." (Question: XP-nSL maf filtering and interpretation of normalisation szpiech/selscan#147 (comment))
2. They further clarified:
   • For XP-nSL: Retaining sites monomorphic in the two compared populations (but polymorphic in the species) is reasonable, as they may influence distance measures.
   • For XP-EHH: Monomorphic sites in both populations only slow down the software (no impact on results), so they can be filtered. (The results are closely related with the monomorphic sites szpiech/selscan#89 (comment))

I want to confirm if this guidance applies to scikit-allel as well:

• Should I completely avoid MAF filtering (e.g., MAF > 0.01/0.05) for both XP-nSL and XP-EHH in scikit-allel?
• For XP-EHH, is filtering only fully monomorphic sites (in both populations) the only acceptable filtering (no MAF-based filtering)?
• For XP-nSL, should I retain even the monomorphic sites in the two compared populations?

Thank you very much for your clarification.

Best regards,
Zheng zhuqing