Related Experiment Videos
Comparison of linear weighting schemes for perfect match and mismatch gene expression levels from microarray data
T Mark Beasley1, Janet K Holt, David B Allison
1Section on Statistical Genetics, Department of Biostatistics, University of Alabama at Birmingham, Birmingham, Alabama 35294, USA.
American Journal of Pharmacogenomics : Genomics-Related Research in Drug Development and Clinical Practice
|June 15, 2005
Summary
The PM-only model offers superior power for analyzing Affymetrix microarray data in most realistic scenarios. This finding supports the growing trend of using PM-only models for gene expression analysis.
Area of Science:
- Bioinformatics
- Computational Biology
- Genomics
Background:
- Affymetrix microarray data analysis employs various methods, including covariate, difference (PM-MM), and PM-only models.
- The covariate model uses a linear function of perfect match (PM) and mismatch (MM) signals.
- The difference model subtracts MM from PM signals, while the PM-only model excludes MM data.
Purpose of the Study:
- To theoretically derive a statistical model for gene expression levels in microarray data.
- To evaluate the performance of different Affymetrix microarray data analytic approaches.
Main Methods:
- Decomposition of correlations within the statistical model.
- Theoretical derivation of a gene expression model under various microarray data conditions.
- Statistical analysis comparing covariate, difference, and PM-only models.
Main Results:
- The covariate model offers flexibility and better reflection of gene expression than the difference model when modeling non-systematic variation.
- The PM-only model demonstrates superior statistical power in most realistic microarray data scenarios.
- Theoretical support is provided for the current trend towards PM-only model utilization.
Conclusions:
- The PM-only model is recommended for Affymetrix microarray data analysis due to its superior power in practical applications.
- The choice of model significantly impacts the accuracy and power of gene expression level estimation.
- This study provides a theoretical framework for understanding and selecting appropriate microarray data analysis methods.