Normalization Benefits Microarray-Based Classification
Author Information
Author(s): Hua Jianping, Balagurunathan Yoganand, Chen Yidong, Lowey James, Bittner Michael L, Xiong Zixiang, Suh Edward, Dougherty Edward R
Primary Institution: Translational Genomics Research Institute
Hypothesis
Normalization procedures can improve classification accuracy in microarray data analysis.
Conclusion
Normalization can significantly benefit classification under difficult experimental conditions, with linear and Lowess regression methods slightly outperforming the offset method.
Supporting Evidence
- Normalization is a common preliminary step in microarray data analysis.
- The study systematically evaluated the effect of normalization on classification.
- Three normalization methods were tested: offset, linear regression, and Lowess regression.
- Seven classification rules were applied to assess the impact of normalization.
Takeaway
When scientists analyze gene data, they need to make sure the data is fair. This study shows that fixing the data helps them make better guesses about what the genes are doing.
Methodology
The study used a model-based approach to generate synthetic gene-expression values and evaluated the effectiveness of three normalization methods and seven classification rules.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website