Skip to main content

Table 3 The rate of risk tier re-classification due to imputation variability using quintile-based cutoffs

From: Genotype imputation and variability in polygenic risk score estimation

Risk tier

Beagle

Eagle+Minimac

SHAPEIT+Minimac

< 20%tile

20–80%tile

> 80%tile

< 20%tile

20–80%tile

> 80%tile

< 20%tile

20–80%tile

> 80%tile

PRSCAD (161)

 < 20%tile

19.89%

0.09%

0%

19.99%

0%

0%

19.27%

0.72%

0%

 20–80%tile

0.04%

59.90%

0.09%

0%

60.01%

0.01%

0.48%

58.88%

0.64%

 > 80%tile

0%

0.10%

19.90%

0%

0.01%

19.98%

0%

0.48%

19.52%

metaGRSCAD (1736608)

 < 20%tile

19.88%

0.11%

0%

19.98%

0.01%

0%

19.47%

0.51%

0%

 20–80%tile

0.29%

59.59%

0.13%

0.01%

60%

0%

0.51%

58.95%

0.54%

 > 80%tile

0%

0.09%

19.92%

0%

0.01%

19.99%

0%

0.50%

19.50%

GPSCAD (6238460)

 < 20%tile

19.85%

0.14%

0%

19.96%

0.02%

0%

19.57%

0.42%

0%

 20–80%tile

0.08%

59.76%

0.17%

0.03%

60%

0%

0.59%

58.93%

0.48%

 > 80%tile

0%

0.13%

19.88%

0%

0%

19.99%

0%

0.56%

19.44%

  1. The rate of re-classification into low risk (< 20th percentile), intermediate risk (20–80th percentile), and high risk (> 80th percentile) due to phasing and imputation variability across all individuals included in this study (1447 EURs and 239 AFRs combined). Columns indicate the average individual-level imputation-based risk tier. Rows indicate the process-level imputation-based risk tier