Improvements to PCA steps & default reference panel

[smlmbrt]

Ideas for making PCA projections more robust

• [ ] Subsetting to a smaller set of PCA eligible variants?
  • HapMap3 (same as bigsnpr)
  • Ancestry informative markers (similar to Hao et al.), such as those in doi:10.3389/fgene.2012.00322
• [x] Avoid variants with high missingness in target dataset (use .vmiss files)
  • Implemented in beta release
• [ ] Add back OCE samples to reference panel, but exclude from empirical calculation of Z-scores due to low number of individuals for comparison.

[smlmbrt]

Previously implemented:

• Merged 1000 Genomes & Human Genome Diversity Project (HGDP):
  • ref: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9900804/
  • data: https://gnomad.broadinstitute.org/news/2020-10-gnomad-v3-1-new-content-methods-annotations-and-data-availability/#the-gnomad-hgdp-and-1000-genomes-callset