A Semantic Network for Modeling Biological Knowledge in Multiple Databases

Jeffrey Stone, Xindong Wu, Marc S. GreenblattCommunications 2015 · DOI

Abstract

We have developed a semantic network of biological terminology to aid in the retrieval and integration of biological information from a variety of disparate information sources. Our semantic network strives to provide a categorization of biological concepts and relationships among these concepts. The semantic network will impart a knowledge structure through which computers can "reason" and draw conclusions about biological data objects and will provide a federated view of the many disparate databases of interest to biologists. In the development of our system, we have included the concepts from several established controlled vocabularies, chief among them being the National Library of Medicine's Unified Medical language System (UMLS). While the UMLS Metathesaurus provides an excellent controlled vocabulary, we have found their semantic network lacking in sufficient detail to be useful as a tool for categorization of biological concepts in databases. Our complete semantic network consists of 183 semantic types and 69 relationships.

Introduction

The rapid growth of data in computational biology and genetics has created a "data management nightmare" for researchers. Traditional search engines rely on keyword processing that lacks biological context, often returning thousands of irrelevant results. For a biologist, a search for a specific protein should ideally understand the relationships between concepts—for example, that a specific gene produces a specific protein with a defined molecular function.

Existing solutions like the National Library of Medicine’s Unified Medical Language System (UMLS) are often too focused on clinical medicine to be useful for pure biological research. Conversely, the UMLS Metathesaurus is too complex for use as a lightweight tool for federating multiple databases. This paper presents a semantic network that bridges this gap, providing a knowledge structure optimized for modeling biological data across disparate sources.

Methodology: Pruning and Grafting

The authors utilized a "prune and graft" approach to build their network. They began with the UMLS Semantic Network and manually removed 16 irrelevant semantic types, such as "manufactured objects." To bolster genomic capabilities, they added 65 new semantic types, 55 of which were adapted from the Gene Ontology (GO).

Crucially, the network introduces 15 new relationship types to handle biological complexity. While most systems rely on hierarchical "is-a" links, this network emphasizes "associate-with" links to model non-hierarchical interactions like "activates" or "disrupts."

The "Associate_with" relationship hierarchy, illustrating non-hierarchical links that allow for complex modeling beyond simple trees.

Results and System Integration

The final network consists of 183 semantic types and 69 relationships. It was tested against several high-profile databases, including the Protein Data Bank (PDB) and Online Mendelian Inheritance in Man (OMIM), to ensure broad coverage.

The network categorizes biological entities into a simplified hierarchy. This structure serves as the "dictionary" for an intelligent digital library system, allowing it to build user profiles and automatically surface relevant data from federated databases.

A portion of the "Entity" semantic types, showing the mix of UMLS (rectangles) and new/external (ovals) types.

Discussion and Future Work

The "Semantic Web" for biology requires domain-specific systems that utilize semantics rather than just keywords. By finding a "middle way" of complexity—finer than the UMLS network but less cumbersome than the full Metathesaurus—the authors provide a tool that enhances precision in data retrieval.

However, the current manual process of pruning and adding nodes is labor-intensive. Additionally, the truncation of the Gene Ontology structure may result in the loss of some highly specific biological nuances. Future work will focus on automating these processes and evolving the network as more diverse database types are added to the digital library.

Summary updated

Abstract

We have developed a semantic network of biological terminology to aid in the retrieval and integration of biological information from a variety of disparate information sources. Our semantic network strives to provide a categorization of biological concepts and relationships among these concepts. The semantic network will impart a knowledge structure through which computers can "reason " and draw conclusions about biological data objects and will provide a federated view of the many disparate databases of interest to biologists. In the development of our system, we have included the concepts from several established controlled vocabularies, chief among them being the National Library of Medicine's Unified Medical language System (UMLS). While the UMLS Metathesaurus provides an excellent controlled vocabulary, we have found their semantic network lacking in sufficient detail to be useful as a tool for categorization of biological concepts in databases. We would like to provide a categorization of concepts that provides finer detail than their semantic network without the considerable size and complexity of their Metathesaurus. Our complete semantic network consists of 183 semantic types and 69 relationships.
BibTeX
@inproceedings{stone2015semantic,
  author = {Jeffrey Stone and Xindong Wu and Marc S. Greenblatt},
  title = {{A Semantic Network for Modeling Biological Knowledge in Multiple Databases}},
  booktitle = {Communications},
  year = {2015},
  doi = {10.58729/1941-6687.1346},
}
Figure