Related Experiment Videos
Inherent bias toward the null hypothesis in conventional multipoint nonparametric linkage analysis
Nicholas J Schork1, Tiffany A Greenwood
1Polymorphism Research Laboratory, Department of Psychiatry, University of California, San Diego, CA 92093-0603, USA. nschork@ucsd.edu
American Journal of Human Genetics
|January 21, 2004
Summary
Traditional linkage analysis methods may inaccurately assign allele-sharing values, potentially biasing results. This bias, stemming from using expected rather than actual allele transmission data, affects complex disease genetic studies.
Area of Science:
- Genetics
- Biostatistics
Background:
- Nonparametric multipoint statistical procedures are used in linkage studies to assign allele-sharing values to relative pairs.
- These methods aim to address challenges like lack of informativity and missing data in genetic marker analysis.
Purpose of the Study:
- To investigate the bias introduced by using expected allele-sharing values in traditional nonparametric linkage analyses.
- To highlight how this bias can skew results toward the null hypothesis, impacting the detection of genetic effects.
Main Methods:
- Analysis of allele-sharing values in relative pairs within a large affected-sibling-pair study for autism.
- Examination of the impact of expected versus actual allele-transmission data on linkage test statistics.
Main Results:
- Traditional methods often assign "expected" allele-sharing values, failing to fully account for data limitations.
- The use of expected values, instead of explicit allele transmission data, can significantly bias linkage tests toward the null hypothesis.
Conclusions:
- The practice of assigning expected allele-sharing values can lead to pronounced bias in nonparametric linkage analyses.
- This bias may explain inconsistencies in applying these methods to complex human traits and diseases, necessitating further research into alternative approaches.