Related Experiment Video
Updated: Mar 9, 2026
JUMPn: A Streamlined Application for Protein Co-Expression Clustering and Network Analysis in Proteomics
Published on: October 19, 2021
The Interaction Network Ontology-supported modeling and mining of complex interactions represented with multiple
Arzucan Özgür1, Junguk Hur2, Yongqun He3,4,5,6
1Department of Computer Engineering, Bogazici University, 34342 Istanbul, Turkey.
Background:
The Interaction Network Ontology (INO) logically represents biological interactions, pathways, and networks. INO has been demonstrated to be valuable in providing a set of structured ontological terms and associated keywords to support literature mining of gene-gene interactions from biomedical literature. However, previous work using INO focused on single keyword matching, while many interactions are represented with two or more interaction keywords used in combination.
Methods:
This paper reports our extension of INO to include combinatory patterns of two or more literature mining keywords co-existing in one sentence to represent specific INO interaction classes. Such keyword combinations and related INO interaction type information could be automatically obtained via SPARQL queries, formatted in Excel format, and used in an INO-supported SciMiner, an in-house literature mining program. We studied the gene interaction sentences from the commonly used benchmark Learning Logic in Language (LLL) dataset and one internally generated vaccine-related dataset to identify and analyze interaction types containing multiple keywords. Patterns obtained from the dependency parse trees of the sentences were used to identify the interaction keywords that are related to each other and collectively represent an interaction type.
Results:
The INO ontology currently has 575 terms including 202 terms under the interaction branch. The relations between the INO interaction types and associated keywords are represented using the INO annotation relations: 'has literature mining keywords' and 'has keyword dependency pattern'. The keyword dependency patterns were generated via running the Stanford Parser to obtain dependency relation types. Out of the 107 interactions in the LLL dataset represented with two-keyword interaction types, 86 were identified by using the direct dependency relations. The LLL dataset contained 34 gene regulation interaction types, each of which associated with multiple keywords. A hierarchical display of these 34 interaction types and their ancestor terms in INO resulted in the identification of specific gene-gene interaction patterns from the LLL dataset. The phenomenon of having multi-keyword interaction types was also frequently observed in the vaccine dataset.
Conclusions:
By modeling and representing multiple textual keywords for interaction types, the extended INO enabled the identification of complex biological gene-gene interactions represented with multiple keywords.
Related Concept Videos
02:26Protein Networks
These interactions can be represented through maps depicting protein-protein interaction networks, represented as nodes and edges. Nodes are circles that are representative of a protein,...
02:26Protein Networks
02:04Protein-protein Interfaces
02:04Protein-Protein Interfaces
01:19Interactions Between Signaling Pathways
Convergence and divergence, and cross-talk between signaling pathways
Two distinct signaling pathways can converge on a single functional unit, which may either be a single protein or a complex of proteins. The response is either functionally distinct or synergistic between the two pathways but different from the response...
02:40Ligand Binding Sites
Protein-ligand interactions are quite specific; even though numerous potential ligands surround a cellular protein at any given time, only a particular ligand can bind to that protein. Moreover, a ligand binds only to a dedicated area on the surface of the protein, known as the...