Clustering rare diseases within an ontology-enriched knowledge graph

Abstract

Objective Identifying sets of rare diseases with shared aspects of etiology and pathophysiology may enable drug repurposing and/or platform based therapeutic development. Toward that aim, we utilized an integrative knowledge graph-based approach to constructing clusters of rare diseases. Materials and Methods Data on 3,242 rare diseases were extracted from the National Center for Advancing Translational Science (NCATS) Genetic and Rare Diseases Information center (GARD) internal data resources. The rare disease data was enriched with additional biomedical data, including gene and phenotype ontologies, biological pathway data and small molecule-target activity data, to create a knowledge graph (KG). Node embeddings were used to convert nodes into vectors upon which k-means clustering was applied. We validated the disease clusters through semantic similarity and feature enrichment analysis. Results A node embedding model was trained on the ontology enriched rare disease KG and k-means clustering was applied to the embedding vectors resulting in 37 disease clusters with a mean size of 87 diseases. We validate the disease clusters quantitatively by looking at semantic similarity of clustered diseases, using the Orphanet Rare Disease Ontology. In addition, the clusters were analyzed for enrichment of associated genes, revealing that the enriched genes within clusters were shown to be highly related. Discussion We demonstrate that node embeddings are an effective method for clustering diseases within a heterogenous KG. Semantically similar diseases and relevant enriched genes have been uncovered within the clusters. Connections between disease clusters and approved or investigational drugs are enumerated for follow-up efforts. Conclusion Our study lays out a method for clustering rare diseases using the graph node embeddings. We develop an easy to maintain pipeline that can be updated when new data on rare diseases emerges. The embeddings themselves can be paired with other representation learning methods for other data types, such as drugs, to address other predictive modeling problems. Detailed subnetwork analysis and in-depth review of individual clusters may lead to translatable findings. Future work will focus on incorporation of additional data sources, with a particular focus on common disease data.

A deep learning transformer model predicts high rates of undiagnosed rare disease in large electronic health systems

“…Deep learning and AI is a rapidly evolving field, and the transformer-based architecture we used for this analysis may not be the optimal way to learn the semantic structure of EHR diagnoses. Recent studies have proposed new ways of derived phenotype embeddings, including extracting them from curated knowledge graphs or from general-purpose large language models pretrained on non-EHR data [68][69][70] . There are also a variety of approaches that have been used to process time series data from EHR in machine learning applications, including neural network models designed for time series data such as recurrent neural networks and using pretrained large language models to process EHR 71,72 .…”

Integrative Rare Disease Biomedical Profile based Network Supporting Drug Repurposing, a case study of Glioblastoma

“…Detailed description of the disease clustering procedure has been described in a separate submission. [27] We extracted 92 subgraphs from the NGKG, each an ego graph[28] of radius of 3 centered on a node containing one of those 92 GBM-related rare diseases.…”

Hypothesis Generation for Rare and Undiagnosed Diseases Through Clustering and Classifying Time-Versioned Biological Ontologies

“…This problem inspired the recent ECoHeN algorithm 36 and is also addressed in proposed alterations to modularity to account for heterogeneous networks 37 ; though this latter approach may still be hindered by the resolution limit inherent to modularity-based approaches 38 . Alternatively, it has been shown how node embeddings can be created and used to represent higher-order patterns in biological knowledge graphs more heterogeneous than those used here 39 .…”