A Robust Measure of Correlation Between Two Genes on a Microarray
Author Information
Author(s): Johanna Hardin, Aya Mitani, Leanne Hicks, Brian Van Koten
Primary Institution: Pomona College
Hypothesis
Can a resistant similarity metric based on Tukey's biweight estimate provide a better measure of correlation for microarray data than traditional methods?
Conclusion
Robust methods, including the biweight correlation, should be used for clustering and gene network analysis of noisy microarray data.
Supporting Evidence
- The resistant metric is shown to be more efficient than Pearson correlation.
- The method provides a systematic gene flagging procedure useful for noisy data.
- Biweight correlation can identify discrepancies in gene relationships that Pearson correlation may miss.
Takeaway
This study shows that using a special way to measure how genes are related can help scientists get better results, especially when the data is messy.
Methodology
The study proposes a resistant similarity metric based on Tukey's biweight estimate of multivariate scale and location, comparing it to Pearson correlation and other methods.
Potential Biases
The study discusses potential biases in traditional correlation methods due to outliers in microarray data.
Limitations
The biweight correlation is computationally intensive and may take longer to compute than other correlation methods.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website