Related Experiment Videos
Power, false discovery rate and Winner's Curse in eQTL studies
Qin Qin Huang1,2, Scott C Ritchie1,3, Marta Brozynska1,3
1Cambridge Baker Systems Genomics Initiative, Baker Heart and Diabetes Institute, 75 Commercial Rd, Melbourne 3004, Victoria, Australia.
Nucleic Acids Research
|September 7, 2018
Summary
Expression quantitative trait loci (eQTL) studies require careful design and analysis. Simulations reveal that common statistical methods inflate false discoveries, especially for low-frequency variants, but BootstrapQTL improves effect size estimation.
Area of Science:
- Genetics
- Statistical Genetics
- Bioinformatics
Background:
- Understanding the genetic basis of gene expression is crucial for interpreting disease associations.
- Key aspects of expression quantitative trait loci (eQTL) study design and analysis require further investigation.
Purpose of the Study:
- To investigate the impact of different eQTL study designs and analysis strategies using simulations.
- To identify optimal methods for controlling false discoveries and estimating effect sizes in eQTL studies.
Main Methods:
- Extensive, empirically driven simulations were performed to evaluate eQTL study designs.
- Various multiple testing correction methods were compared, including false discovery rate (FDR) control and hierarchical procedures.
- A novel bootstrap method, BootstrapQTL, was developed for improved effect size estimation.
Main Results:
- False discoveries of eGenes were inflated with standard FDR control, particularly for low-frequency single nucleotide polymorphisms (SNPs) and small sample sizes.
- Power was low and FDR was inflated for eGenes with low-frequency causal SNPs (<10% minor allele frequency) in small samples (100 individuals).
- The 'Winner's Curse' (overestimation of effect sizes) was prevalent in low to moderate power settings, but BootstrapQTL provided more accurate estimates.
Conclusions:
- Standard multiple testing correction methods can lead to inflated false discoveries in eQTL studies, especially for rare variants.
- Careful consideration of study design, sample size, and allele frequencies is essential for robust eQTL analysis.
- BootstrapQTL offers a promising approach for mitigating the Winner's Curse and improving effect size accuracy in eQTL studies.