Figure 2.

Block‐specific posterior mean numbers of selected single nucleotide polymorphisms by the JAM method in each of the three meta‐GWAS case studies. The vertical spread indicates the variation in block‐specific adapted sparsities from step one of our proposed framework. Note that the global average varies across the case studies owing to different optimal λs, which control the global sparsity. For CAD and schizophrenia, summary statistics were available from considerably larger training datasets, allowing the estimation of many more small effects (see Table 2). CAD, coronary artery disease