Representing glycophenotypes: semantic unification of glycobiology resources for disease discovery.

Jean-Philippe F Gourdine,Julie A Mcmurry,Nicolas Matentzoglu,Melissa A Haendel,Matthew H Brush,Kent Shefchek,Xingmin Aaron Zhang,Monica C Munoz-Torres,Peter N Robinson,Nicole A Vasilevsky,Sebastian Köhler

doi:10.1093/database/baz114

Abstract

While abnormalities related to carbohydrates (glycans) are frequent for patients with rare and undiagnosed diseases as well as in many common diseases, these glycan-related phenotypes (glycophenotypes) are not well represented in knowledge bases (KBs). If glycan-related diseases were more robustly represented and curated with glycophenotypes, these could be used for molecular phenotyping to help to realize the goals of precision medicine. Diagnosis of rare diseases by computational cross-species comparison of genotype–phenotype data has been facilitated by leveraging ontological representations of clinical phenotypes, using Human Phenotype Ontology (HPO), and model organism ontologies such as Mammalian Phenotype Ontology (MP) in the context of the Monarch Initiative. In this article, we discuss the importance and complexity of glycobiology and review the structure of glycan-related content from existing KBs and biological ontologies. We show how semantically structuring knowledge about the annotation of glycophenotypes could enhance disease diagnosis, and propose a solution to integrate glycophenotypes and related diseases into the Unified Phenotype Ontology (uPheno), HPO, Monarch and other KBs. We encourage the community to practice good identifier hygiene for glycans in support of semantic analysis, and clinicians to add glycomics to their diagnostic analyses of rare diseases.

Highlights

From antiquity to present days, clinicians have described diseases with phenotypic features mostly in a free-text representation—from ancient Egyptians using papyrus [1] to today’s disease descriptions in textbooks, publications or medical records
The Open Biological and Biomedical Ontology (OBO) Foundry has developed standards for logically well-formed and interoperable ontologies respectful of the representations of biological reality [4]. These ontologies are often used in knowledge bases (KBs) and DBs to semantically structure information and allow for computational classification and inferencing across data
We focused on diseases and/or glycans by highlighting features, applications, uses and challenges in order to provide potential resources for representing glycophenotypes in a compatible way with the Human Phenotype Ontology (HPO), and applying them toward phenotype-based patient diagnosis and disease-gene discovery

Summary

Introduction

From antiquity to present days, clinicians have described diseases with phenotypic features mostly in a free-text representation—from ancient Egyptians using papyrus [1] to today’s disease descriptions in textbooks, publications or medical records.

Results

Conclusion