Improving Gene Expression Analysis with Stratified False Discovery Rate Control
Author Information
Author(s): Huang Baisong, Rangrej Jagadish, Paterson Andrew D, Sun Lei
Primary Institution: Research Institute, Hospital for Sick Children
Hypothesis
Can the stratified false-discovery rate control method improve the performance of genome-wide linkage analyses compared to traditional methods?
Conclusion
The stratified false-discovery rate method outperformed the traditional aggregated approach in identifying true positives in gene expression data.
Supporting Evidence
- The stratified method provided 57 more true positives while controlling the FDR at the same rate as the aggregated approach.
- Among the 9,069,390 tests, more than half had p-values equal to 1, indicating a significant association with H2 = 0.
- The cis stratum contained a relatively higher proportion of true signals than the trans stratum.
Takeaway
This study shows that using special methods to analyze gene data can help find more true signals and reduce mistakes.
Methodology
The study performed eQTL analysis using MERLIN and compared the stratified FDR method with traditional aggregated approaches under three frameworks.
Limitations
The stratified method did not outperform the aggregated approach by as large a margin as expected due to the overwhelming number of trans regulators.
Statistical Information
P-Value
1
Statistical Significance
p<0.0001
Want to read the original?
Access the complete publication on the publisher's website