Linking diseases and phenotypes for clinical usage

Background: In recent years a large volume of clinical genomics data has become available due to rapid advances in sequencing technologies. Efficient exploitation of genomics data requires linkage to patient phenotype profiles. Current resources, providing disease-phenotype associations, are not comprehensive, and they often do not cover all of the disease terminologies, particularly ICD-10, which is still the primary terminology used in clinical settings. Methods: We used a text-mining method that utilizes semantic relations in phenotype ontologies, and applied statistical methods to extract associations between diseases in ICD-10 and phenotype ontology classes from the literature. In addition, we present a semi-automatic way to curate ICD-10:phenotype associations from existing resources containing known associations. Results: We created two datasets linking diseases to their phenotypes based on each strategy. We extensively validated these datasets based on comparison to gold standard manually created disease--phenotype associations, based on their similarity to disease--phenotype associations found in databases, and based on how well they can be used to recover gene--disease associations using phenotype similarity. We find that our text mining method can produce phenotypes that are useful but often too specific or too general. However, the dataset generated from integrating multiple knowledgebases is more granular