Estimating Gene Product Similarity Using a Graph Method
Author Information
Author(s): Alvarez Marco A, Qi Xiaojun, Yan Changhui
Primary Institution: Utah State University
Hypothesis
Can a shortest-path graph kernel method provide a better estimation of semantic similarity between gene products using only the Gene Ontology?
Conclusion
The shortest-path graph kernel method provides a comparable performance to existing methods while relying solely on the Gene Ontology.
Supporting Evidence
- The spgk method showed a high correlation with functional similarities across three ontologies.
- Statistical tests indicated significant improvements in performance metrics when compared to existing methods.
- The method is computationally efficient, operating in polynomial time.
Takeaway
This study created a new way to measure how similar genes are by using a special graph method that only looks at the Gene Ontology, which is like a big dictionary for genes.
Methodology
The study used a shortest-path graph kernel method to calculate semantic similarity between gene products based on their representation as graphs derived from the Gene Ontology.
Potential Biases
The method may not account for different biological meanings associated with various types of relationships in the GO.
Limitations
The method currently does not utilize the rich text definitions associated with GO terms for node comparison.
Participant Demographics
The study evaluated 100 proteins with the highest number of GO annotations.
Statistical Information
P-Value
p<0.001 for resolution, p=0.0384 for EC similarity, p=0.2266 for Pfam similarity
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website