Evaluating Protein Semantic Similarity Measures
Author Information
Author(s): Pesquita Catia, Faria Daniel, Bastos Hugo, Ferreira António E, Falcão André O, Couto Francisco M
Primary Institution: Universidade de Lisboa
Hypothesis
Which semantic similarity measure performs better with Gene Ontology annotations?
Conclusion
The simGIC measure showed the highest resolution in capturing the relationship between functional and sequence similarity.
Supporting Evidence
- The study systematically evaluated fourteen semantic similarity measures.
- Resnik's measure was found to have the lowest bias and highest resolution.
- The best-match average approach was the most effective for combining term similarities.
- Electronic annotations were shown to influence the results significantly.
Takeaway
This study looked at different ways to compare proteins based on their functions, and found that one method, simGIC, works best.
Methodology
A systematic evaluation of semantic similarity measures was conducted using sequence similarity as a performance metric.
Potential Biases
Potential bias from using electronic annotations that may not always be accurate.
Limitations
The study's findings may be influenced by data circularity due to the use of electronic annotations.
Participant Demographics
A subset of 22,067 Swiss-Prot proteins annotated with molecular function GO terms.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website