A comprehensive evaluation of SAM, the SAM R-package and a simple modification to improve its performance
2007

Evaluating the SAM R-Package for Gene Expression Analysis

Sample size: 5000 publication Evidence: moderate

Author Information

Author(s): Zhang Shunpu

Primary Institution: Department of Statistics, University of Nebraska Lincoln

Hypothesis

The study aims to evaluate the performance of the SAM method and its R-package sam2.20 in controlling the false discovery rate (FDR) in microarray data analysis.

Conclusion

The study concludes that while sam2.20 improves upon SAM, it still does not adequately control the FDR, and a symmetric cutoff method may perform better under certain conditions.

Supporting Evidence

  • The study identifies discrepancies between SAM and sam2.20 in FDR estimation methods.
  • Numerical simulations show that the symmetric cutoff method outperforms sam2.20 under certain conditions.
  • The proposed modification to sam2.20 improves FDR control but does not eliminate all issues.

Takeaway

This study looks at how well a method for finding important genes works, and it finds that even the improved version still makes mistakes.

Methodology

The study involved simulations comparing the performance of SAM and sam2.20 against a symmetric cutoff method in detecting differentially expressed genes.

Potential Biases

The use of the same statistic for both test and null distributions may lead to biased estimates of FDR.

Limitations

The study acknowledges that both methods may significantly overestimate the true FDR and that the performance can vary based on the distribution of induced and repressed genes.

Digital Object Identifier (DOI)

10.1186/1471-2105-8-230

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication