Figure 4

Histogram of the ranks of genes predicted for validation sets across the SNP/gene search spaces: CPS predictions against the (A) Known disease gene set and (B) WTCCC-implicated gene candidates; CMP predictions against the (C) Known disease gene set and (D) WTCCC-implicated gene candidates. The data sets are shaded based on the significance of the underlying SNP set: yellow for HS, orange for MHS, purple for MWS and dark purple for WS. In each set predictions made using known disease genes as seeds are shown on a grey background and ab initio predictions are shown on the white background. The graph shows that the priority assigned to a candidate gene prediction by Gentrepid agrees well with the significance of the underlying SNP. Predictions based on the most significant HS SNPs are clustered at the top of the figure (ranked first), showing prioritisation by CPS is effective. CMP, on the other hand, effectively ranks a handful of predictions made in ab initio mode when judged against WTCCC candidates. The majority of CMP predictions have not been previously detected. Ranks are displayed up until 10^th place.