Additional file 2: Figure S2. of A visual and curatorial approach to clinical variant prioritization and disease gene discovery in genome-wide diagnostics

Performance of global map visual projection. Semantic space signal-to-noise ratios of known disease classes are recovered in MDS visual space. Signal is computed as the mean semantic similarity between all pairs of diseases within a known (A) HDN or (B) OMIM Phenotypic Series class (gray bars). Noise is computed as the mean similarity between all pairs of diseases in each class C and those not in C (black line). Scores are relativized to the highest within-class average. Signal-to-noise ratios were consistently above one, indicating retention of pre-MDS semantic space relationships in post-MDS visual space. (PDF 259 kb)