Figure 5.

Comparison of association statistics between annotated pairs and not-annotated pairs

(A) Boxplot of the Chi-squared, Jaccard index, and odds ratio distribution (log scale) calculated in the OARD for annotated disease-phenotype pairs and not-annotated disease-phenotype pairs.

(B) Distribution of the p value of the concepts obtained from the Wilcoxon test by comparing its annotated pairs and not-annotated pairs. The dashed line is p value = 0.05.

(C) Distribution of the mean difference of the statistics between annotated pair and not-annotated pair of a given disease for different evidence levels. IEA, annotations that have been extracted by parsing the clinical features sections of the OMIM; PCS, annotations that have been extracted from articles in the medical literature; TAS, annotations that have a “traceable author statement”, usually reviews or disease entries (e.g., OMIM) that only refers to the original publication.

(D) Distribution of the mean difference of the statistics between annotated pair and not-annotated pair of a given disease for different levels of frequency that a phenotype was found in the given diseases. Very rare, in 1% to 4% of the cases; occasional: in 5% to 29% of the cases; frequent, in 30% to 79% of the cases; very frequent, in 80% to 99% of the cases.