Related Experiment Videos
A Monte Carlo procedure for two-stage tests with correlated data.
1Section of Medical Genetics, Department of Medicine, Duke University Medical Center, Durham, NC 27710, USA. emartin@chg.mc.duke.edu
Genetic Epidemiology
|December 22, 1999
Summary
This study introduces a Monte Carlo method to accurately assess disease locus associations using a two-stage approach. It corrects for bias when combining case-control and family data in the first stage, improving genetic analysis power.
Area of Science:
- Genetics
- Statistical genomics
- Genetic epidemiology
Background:
- Two-stage strategies combining case-control and family-based association tests are used for mapping disease loci.
- First-stage association tests require high power to detect true signals, as only positive results proceed to the second stage.
- Combining case-control and family data in the first stage can increase power but introduces correlation, potentially biasing significance levels.
Purpose of the Study:
- To propose a statistical method that accurately accounts for correlation between first- and second-stage tests in a two-stage disease locus mapping strategy.
- To provide a correct significance level for the second-stage test when family and case-control data are combined in the first stage.
- To discuss the application of this two-stage procedure in genome scans, specifically for data from the Genetic Analysis Workshop 9 study.
Main Methods:
- Development of a Monte Carlo simulation method to address the correlation between combined first-stage (case-control and family data) and second-stage (family-based) tests.
- Evaluation of the proposed method's ability to provide accurate significance levels for the second-stage association test.
- Application and discussion of the method in the context of genome-wide association studies.
Main Results:
- The proposed Monte Carlo method effectively accounts for the correlation introduced by combining data in the first stage.
- This method provides the correct significance level for the second-stage family-based test, mitigating bias.
- The approach enhances the reliability of results from two-stage genetic association studies.
Conclusions:
- A novel Monte Carlo method is presented for accurate statistical inference in two-stage genetic association studies.
- This method is crucial for maintaining appropriate significance levels when family and case-control data are integrated in the initial stage.
- The proposed approach improves the robustness of disease locus mapping, particularly in genome scan contexts.