Figure 2

Construction and evaluation of models for (prior) probability of association given the functional annotation data. The purple arrows represent model construction (‘training’), while the green arrows represent evaluation of the models. Construction of the training set, validation set and functional annotation database are depicted in Additional file 1: Figure S10 and described in Methods. The training data were used to construct a series of models, each distinguished by the coefficients (or ‘weights’) it assigns to the various annotation variables. We chose the best among these by comparing their predictions in the validation set using the concordance index.