Phenotype to Gene Search

Phenotype to gene search finds similar phenotypes and diseases across various ontology sources, independent of the underlying vocabulary in each source. If an equivalent concept does not exist across the sources, TruSight Software calculates the distance between nodes in the ontological hierarchies and assigns a score from 0–1, where:

•

Values closer to 0 indicate that the concepts are more equivalent. A value of 0 indicates that the concepts are the same.

•

Values closer to 1 indicate that the concepts are more dissimilar. A value of 1 indicates that the concepts can only be connected at the root node and are therefore excluded from query results.

The determination of distance accounts for the fact that sibling concepts on leaf nodes (eg, hypertrophic cardiomyopathy and dilated cardiomyopathy) are more closely related to each other than siblings close to the root (eg, abnormal vascular morphology and abnormal heart morphology).

Confidence scores are calculated using the following rules:

•

Expert-curated data from OMIM, HPO, Phenopedia, and ClinVar are assigned a high confidence score.

•

High, moderate, or low confidence scores are converted from GeL PanelApp strong medium, and low scores, respectively.

•

GTR confidence scores are based on information content metrics, which measure the specificity of a genetic test for a particular phenotype and a gene.

•

GeneRIF associations, which are derived using data mining, are assigned medium confidence.