Improving Gene Prediction with a Generalized Hidden Markov Model
Author Information
Author(s): Stanke Mario, Schöffmann Oliver, Morgenstern Burkhard, Waack Stephan
Primary Institution: Universität Göttingen
Hypothesis
Can integrating extrinsic evidence improve gene prediction accuracy in eukaryotes?
Conclusion
Using a Generalized Hidden Markov Model that incorporates both intrinsic and extrinsic evidence significantly enhances gene prediction accuracy.
Supporting Evidence
- The new method, AUGUSTUS+, predicted 89% of the exons in human chromosome 22 correctly.
- AUGUSTUS+ outperformed other gene prediction programs in accuracy.
- Combining intrinsic and extrinsic information improved the specificity of gene predictions.
Takeaway
This study shows that using extra information from databases can help computers better guess where genes are in DNA.
Methodology
The study developed a Generalized Hidden Markov Model that integrates hints from external sources like EST and protein databases to improve gene prediction.
Potential Biases
Potential biases may arise from the reliance on external databases which may not be comprehensive.
Limitations
The model may not accurately predict non-coding regions and relies on the quality of external hints.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website