Modeling and integration of N-glycan biomarkers in a comprehensive biomarker data model

Abstract

Molecular biomarkers measure discrete components of biological processes that can contribute to disorders when impaired. Great interest exists in discovering early cancer biomarkers to improve outcomes. Biomarkers represented in a standardized data model, integrated with multi-omics data, may improve the understanding and use of novel biomarkers such as glycans and glycoconjugates. Among altered components in tumorigenesis, N-glycans exhibit substantial biomarker potential, when analyzed with their protein carriers. However, such data are distributed across publications and databases of diverse formats, which hamper their use in research and clinical application. Mass spectrometry measures of 50 N-glycans on 7 serum proteins in liver disease were integrated (as a panel) into a cancer biomarker data model, providing a unique identifier, standard nomenclature, links to glycan resources, and accession and ontology annotations to standard protein, gene, disease, and biomarker information. Data provenance was documented with a standardized United States Food and Drug Administration-supported BioCompute Object. Using the biomarker data model allows the capture of granular information, such as glycans with different levels of abundance in cirrhosis, hepatocellular carcinoma, and transplant groups. Such representation in a standardized data model harmonizes glycomics data in a unified framework, making glycan-protein biomarker data exploration more available to investigators and to other data resources. The biomarker data model we describe can be used by researchers to describe their novel glycan and glycoconjugate biomarkers; it can integrate N-glycan biomarker data with multi-source biomedical data and can foster discovery and insight within a unified data framework for glycan biomarker representation, thereby making the data FAIR (Findable, Accessible, Interoperable, Reusable) (https://www.go-fair.org/fair-principles/).

Keywords: N-linked glycans; cancer biomarker panel; data integration; glyco-informatics; liver disease.

Fig. 1

Conceptual diagram depicting the majority of biomarker-related data types (dark boxes) represented in the N-glycan biomarker data model and named relations (arrows, gray boxes) existing among some data types. ac, accession; G2C, GlyTouCan; m/z, mass-to-charge ratio; x_ref, cross-reference. Figure adapted from Gogate et al. (2021).

Fig. 2

N-glycan biomarker data panel for the Glyco-typer liver disease assay. Data produced by assay of specific N-glycans on serum glycoproteins of liver disease cohorts are represented on a Cancer Biomarkers home page (https://data.oncomx.org/cancerbiomarkers) as a single data element (biomarker panel). The panel encapsulates key high-level descriptors of the aggregate data (left column), as well as details and specific links to standard nomenclatures, glycan resources, ontologies, and a link to the data model of the set (https://data.oncomx.org/OMX_000059) (right column). dHex, 6-deoxy-hexose; DOID, Disease Ontology Identifier; Hex, Hexose; HexNAc, N-acetylhexosamine; ID, identifier; NeuAc, N-acetylneuraminic acid; NeuGc, N-glycolylneuraminic acid; PMID, PubMed identifier; UN, Uber-Anatomy Ontology; x_ref, cross-reference.