Consistency

---

---

Purpose

Collaborators: Bill Baumgartner, Nicole Vasilevsky

This page describes the validation work performed in order to demonstrate the logical consistency of the OMOP2OBO mappings. For additional information on how we create semantic representations of the OMOP2OBO mappings, please see this wiki page.

The experiment described below was designed to assess the semantic representation of the mappings.

Table of Contents

• Experiments
• Results

---

---

Experiment

---

Experiments were designed to evaluate the logical consistency of each mapping with respect to its specific ontology. Although we have mapped multiple ontologies to each clinical domain (table shown below), this experiment aimed to evaluate the consistency of each of the mappings with respect to each contributing ontology.

Clinical Domains and Ontologies

Clinical Domain	Ontologies
Conditions	HPO, MONDO
Drug Ingredients	CHEBI, PRO, NCBITaxon, VO
Measurements	HPO, UBERON, CL, CHEBI, NCBITaxon, PRO

For each ontology, a new ontology class for all mappings involving 2 or more ontology concepts (since single concepts already exist within each ontology) will be created. Once all mappings for a given ontology are generated (using the patterns described in the Semantic Mapping Representation section), reasoners will be run and the output evaluated as described below.

Reasoner(s):

• Use Hermit first and if that does not work, then ELK
• Run reasoner on each ontology before and adding new concepts

Reasoner Output and Statistics:
For each run, the following results will be recorded:

• Whether or not the reasoner finished. If not, record the reasons why and how the issues were identified
• The number of new inferences generated that include the mappings

Results

---

Currently, the code to automatically generate the mappings has only been tested for condition concepts. This enables the examination of the mapping definitions for HPO and MONDO.

For both HPO and MONDO, the ELK reasoner successfully completed with no errors.

• For MONDO, adding the OMOP2OBO mappings resulted in an increase of 243,917 triples and 41,155 classes. Running the ELK reasoner resulted in an increase of 2,243,134 triples
• For HPO, adding the OMOP2OBO mappings resulted in an increase of 239,876 triples and 37,114 classes. Running the ELK reasoner resulted in an increase of 1,425,135 triples

---