Table 1 Comparison of best known odds ratios in the literature (Literature) to the odds ratios calculated from UK BioBank data presented here (New).

From: Genomic Prediction of 16 Complex Disease Risks Including Heart Attack, Diabetes, Breast and Prostate Cancer

ConditionOdds Ratio
PGS %LiteratureNew99% Predicted
Atrial Fibrillation>90%\({2.74}_{-0.22}^{+0.19}\)* 13\({2.81}_{-0.24}^{+0.24}\)\({10.8}_{-1.6}^{+2.1}\)
Basal Cell Carcinoma>96%\({2.64}_{-0.36}^{+0.36}\)\({3.8}_{-0.54}^{+0.88}\)
Breast Cancer>96%\({2.36}_{-0.16}^{+0.18}\)* 13\({1.799}_{-0.27}^{+0.27}\)\({2.5}_{-0.10}^{+0.14}\)
Heart Attack>96%\({2.25}_{-0.37}^{+0.37}\)\({2.7}_{-0.28}^{+0.52}\)
High Cholesterol>96%\({2.54}_{-0.27}^{+0.27}\)\({2.29}_{-0.38}^{+0.58}\)
Malignant Melanoma1σ shift\({1.36}_{-0.15}^{+0.16}\)61\({1.35}_{-0.26}^{+0.26}\)\({4.28}_{-0.98}^{+0.89}\)
Prostate Cancer>75%/<25%\({3.3}_{-0.6}^{+0.6}\)* 62\({1.58}_{-0.34}^{+0.34}\)\({4.6}_{-0.25}^{+0.33}\)
Testicular Cancer>96%\({1.73}_{-0.97}^{+0.97}\)\({1.13}_{-0.42}^{+1.54}\)
Type 1 Diabetes>95%22.8* 63\({4.22}_{-0.44}^{+0.44}\)\({13.73}_{-0.79}^{+1.16}\)
Type 2 Diabetes>90%\({2.52}_{-0.17}^{+0.19}\)* 13\({2.04}_{-0.05}^{+0.05}\)\({2.81}_{-0.27}^{+0.27}\)
  1. Comparison was either made at the largest possible PGS common to the two sets, or using whatever definition of odds ratio was used in the literature (PGS %). Additionally we indicate what we predict the odds ratio will be for those with 99% scores or above (99% Predicted column). These predictions are found by assuming the data was drawn from Gaussian distributions. We confine our references to the literature to specifically genetic or polygenic risk score determination of odds ratios. Other biological risk factors could, in the future, be combined with genetic risk to generate even better prediction. Further details about the literature are found in Section E. We focus here on purely genetic predictors. For many traits we were unaware of previous odds ratio estimates based on a purely polygenic score. For those we were aware of we listed the largest odds ratio in the chart above. *These predictors include a regression on non-genetic biological information. This article appeared on the BioRxiv shortly before our manuscript and we were originally unaware of the results.