Comparing segmentations by applying randomization techniques
2007
Evaluating Genomic Segmentation Techniques
publication
Evidence: moderate
Author Information
Author(s): Haiminen Niina, Mannila Heikki, Terzi Evimaria
Primary Institution: University of Helsinki
Hypothesis
Can randomization techniques effectively evaluate the quality of genomic segmentations?
Conclusion
The study introduces a framework for evaluating segmentation quality and demonstrates its application on genomic structures.
Supporting Evidence
- Randomization tests showed that some segmentations were significantly closer to the true segmentation than random ones.
- The framework can evaluate segmentations based on different biological features.
- Conditional entropy was used to measure the similarity between segmentations.
Takeaway
This study helps scientists understand how to check if their methods for dividing DNA into segments are good or just lucky.
Methodology
The study uses randomization techniques to compare segmentations of genomic sequences based on conditional entropy.
Limitations
The method may not apply well to segmentations with very different numbers of segments.
Statistical Information
P-Value
0.0018
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website