Performance of the different modules on the GWAS data compared to randomly generated data. Each point represents the log of the ratio of predictions for a phenotype-specific data set compared to a similar randomly generated SNP set. A positive ratio indicates better performance by the system. For the HS and MHS sets, the system performs better under most SNP/gene approaches, with the MHS set performing the best. In the MWS and WS sets, performance depends on the SNP/gene approach used to map the search spaces, with the NN gene set outperforming the BY sets. The nearest, adjacent and 0.1 Mbp approaches have the best performance.