Assays |
Typically captured using inexpensive genotyping microarrays |
Often requires NGS, especially for detecting extremely rare/novel variants |
Number of variants tested |
Often single variant based (e.g., GWAS) |
Often multiple variants based due to low power of single-variant methods |
Population structure |
Confounding can be adequately controlled using PCA or mixed models |
Rare variants are likely more recent and reflect finer subpopulations. May need either more PCs or specifically designed methods |
Null distributions of test statistics |
Ordinary asymptotic distributions work well |
Null distributions are often complex mixtures and more sophisticated methods may be necessary |
Use of annotations |
Statistical test for each variant is often performed without relying on annotations |
Due to the large number of rare variants in a region, annotations are often used to filter rare variants |
Interpretation |
Due to potential LD, single-variant associations may be tag-SNPs |
May be unclear which RVs are “driving” a significant RV association result using aggregative testing, especially those considering both directions |