Meta-analysis based on control of false discovery rate: combining yeast ChIP-chip datasets

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• High-throughput microarray technology enables the measurement of thousands of gene expression levels simultaneously.
• Combining significance values across experiments and controlling false positives are critical challenges in microarray bioinformatics.
• Current methods often address these issues independently, limiting their effectiveness.

Purpose of the Study:

• To introduce a novel, interdependent method for combining microarray significance values.
• To enhance statistical power while maintaining control over the false discovery rate (FDR).
• To provide a comprehensive solution for quality control and speed in meta-analysis.

Main Methods:

• A generalized truncated product method is employed to combine significance values.
• Significance values are filtered based on experiment-specific false discovery thresholds.
• The method integrates FDR control, high power, quality control, and speed-up mechanisms.

Main Results:

• The novel method (ESP) successfully combines significance values, controlling the overall FDR.
• It demonstrates high statistical power in identifying significant features across experiments.
• Application to ChIP-chip data identified regulatory sequences for yeast transcription factors with controlled FDR.

Conclusions:

• The developed method offers an interdependent approach to microarray meta-analysis.
• It effectively addresses the dual needs of high power and FDR control.
• This computational approach is crucial for analyzing accumulating genome-wide datasets.