Improving Gene Function Classification with Bayesian Models
Author Information
Author(s): Shahbaba Babak, Neal Radford M
Primary Institution: University of Toronto
Hypothesis
Can gene function annotation be improved using a classification scheme that incorporates hierarchical organization of functional classes?
Conclusion
The study demonstrates that gene function can be predicted with higher accuracy using Bayesian models that utilize hierarchical information.
Supporting Evidence
- The corMNL model outperformed other models in predicting gene functions.
- Using hierarchical information significantly improved predictive accuracy.
- The study utilized a dataset of 2122 Open Reading Frames (ORFs) from the E. coli genome.
Takeaway
This study shows that using smart models can help us guess what genes do better than before, especially by looking at how they are related.
Methodology
The study applied three Bayesian models (MNL, treeMNL, and corMNL) to predict gene functions based on phylogenetic descriptors, sequence attributes, and predicted secondary structures.
Limitations
The study used a dataset from 2001, which may not reflect the most current gene functions.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website