Mixture Model Approach to Multiple Testing in Genetic Analysis
Author Information
Author(s): Dalmasso Cyril, Pickrell Joseph, Tuefferd Marianne, Génin Emmanuelle, Bourgain Catherine, Broët Philippe
Primary Institution: Universite Paris-Sud, INSERM UMR-S 535, The University of Chicago
Hypothesis
Can a finite mixture model effectively estimate local false discovery rates in genetic studies?
Conclusion
The proposed mixture model provides a more accurate estimation of the null distribution, improving the reliability of false discovery rate estimates in genetic studies.
Supporting Evidence
- The mixture model allows for empirical estimation of the null distribution.
- The estimated null distribution parameters differed significantly from theoretical values.
- The approach reduces the computational burden compared to traditional simulation methods.
- Using the empirical null distribution led to fewer significant results than the theoretical one.
Takeaway
This study shows a new way to figure out if the results from genetic tests are real or just mistakes, making it easier to trust the findings.
Methodology
The study used a two-component mixture model to estimate local false discovery rates, false discovery rates, and false non-discovery rates in variance-component linkage analysis.
Limitations
The approach may not be directly comparable to traditional methods due to different selection criteria.
Participant Demographics
The study involved 194 individuals from 14 three-generation families.
Want to read the original?
Access the complete publication on the publisher's website