Fig. 2From: SoftPanel: a website for grouping diseases and related disorders for generation of customized panelsROC curves of phenotype similarity matrices with different methods of weighting. ROC analysis with the two benchmark datasets (a: Phenotypic Series, b: Linked OMIM Record Pairs) showed that global weighting was superior to other forms of weighting. The range of false positive rates was restricted to (0, 0.1) in order to highlight the differences between each curve more clearlyBack to article page