LD reference panels - All of Us and UK Biobank

All of Us Reference Panels

LD reference panels construction

• Select genetic variants that are available in the All of Us (CDRv7, Controlled Tier) array genotype dataset.
• The following variant filtering/removing criteria were applied: (a) located on sex chromosomes; (b) missing value frequency > 0.03; and (c) MAF < 0.01.
• LD blocks were defined according to the LD block definition in PRS-CSx [Ref] (the start and end base pair position of the blocks of the LD reference panels used in PRS-CSx).
• LD blocks were saved in matrices with the "dgCMatrix" class and in "RDS" file format, together compressed with map files including the variant information.

Sample size

• African American: 62,331
• Hispanic: 59,640

Number of variants

• African American: 993,783
• Hispanic: 875,221

UK Biobank Reference Panels

LD reference panels construction

• Select genetic variants that are available in both 1000 Genomes and the UK Biobank genotyped dataset.
• The following variant filtering/removing criteria were applied: (a) located on sex chromosomes; (b) missing value frequency > 0.02; and (c) MAF < 0.05.
• LD blocks were saved in matrices with the "dgCMatrix" class and in "RDS" file format, together compressed with map files including the variant information.

Sample size

• AFR:7,507
• AMR: 687

Number of variants

• AFR: 1,234,911
• AMR: 1,183,556