Table 3 Comparison of GWAS performed on three phenotypes of interest and replicability of the significantly detected loci

From: Deep learning-based phenotype imputation on population-scale biobank data increases genetic discoveries

Replication within UKBB (imputed versus observed)

 

Imputed–Observed

Phenotype

Effect

Sig

Bilirubin

17 of 17

17 of 17

LifetimeMDD

14 of 14

14 of 23

Cannabis ever taken

7 of 7

7 of 11

Replication of UKBB associations (observed and imputed) in external datasets

 

Observed–External

Imputed–External

Phenotype

Study

Effect

Sig

Effect

Sig

Bilirubin

ref. 25

14 of 15

15 of 32

8 of 9

9 of 14

LifetimeMDD

ref. 26

0 of 0

0 of 1

13 of 13

13 of 18

LifetimeMDD

ref. 17

0 of 0

0 of 1

16 of 16

16 of 20

Cannabis ever taken

ref. 27

0 of 0

0 of 0

1 of 1

1 of 9

  1. The significantly detected loci in one study were compared with another in terms of how many loci match in effect direction of the SNP (Effect) where effect size was distinct from zero in the replication study (P < 0.05), and how many were marginally significant (Sig) overall (P < 0.05). Significance was determined using a two-sided t-test. Comparisons were performed between imputed and originally observed phenotypes in UKBB first (upper) and then between the associations detected in the former two and association studies external to UKBB (lower). Denominator indicates the total number of SNPs present in the compared study and matching the criteria for comparison.
Back to article page