Performance comparison summary for test pedigree in Figure 1.

denotes the lifetime morbid risk, and denotes the heritability of liability. The next three columns provide discriminative accuracies (measured in terms of ) for complete family history, restricted family history, and currently known SNP associations. The last three columns show the corresponding estimated proportion of heritability explained by each model (for family history models, this is taken to be the proportion of heritability that a SNP-based model would need to explain in order to obtain the given ). Bolded, italicized rows indicate diseases for which current SNP-based models outperform the complete family history model; bolded, non-italicized rows indicate diseases for which current SNP-based models outperform the restricted but not the complete model. Diseases were selected based on availability of disease frequency and heritability estimates; references for values of , , and (for SNP-based models) are provided in Text S2 and Table 3. Note that the performance of SNP-based models shown here reflects only currently known genetic factors for European populations and will change as more associations are discovered.