Related Experiment Videos
Haplotype frequency estimation error analysis in the presence of missing genotype data
Enda D Kelly1, Fabian Sievers, Ross McManus
1Hitachi Dublin Lab,, Hitachi Europe Ltd., O'Reilly Institute, Trinity College, Dublin 2, Ireland. enda.kelly@hitachi-eu.com
BMC Bioinformatics
|December 3, 2004
Summary
Haplotype analysis using the Expectation-Maximisation (EM) algorithm remains accurate even with up to 30% missing data. This finding supports treating ambiguous data points as unknown in population and linkage disequilibrium studies.
Area of Science:
- Genetics
- Bioinformatics
- Computational Biology
Background:
- Haplotype analysis is crucial for population studies, linkage disequilibrium, and candidate gene analysis.
- Expectation-Maximisation (EM) algorithms are widely used for haplotype phasing with unknown data.
- The impact of missing allelic data on EM-based methods has been a subject of recent speculation.
Purpose of the Study:
- To evaluate the performance of EM-based haplotype analysis methods in the presence of missing data.
- To develop and test a modified EM program that accommodates missing data.
- To assess the accuracy of confidence intervals using non-parametric bootstrapping.
Main Methods:
- Development of a modified Expectation-Maximisation (EM) program to handle missing allelic data.
- Incorporation of non-parametric bootstrapping for calculating confidence intervals.
- Analysis of various datasets with randomly introduced missing data (up to 30%).
Main Results:
- EM-based haplotype analysis tolerates up to 30% missing data in both biallelic and multiallelic datasets.
- Moderate to strong linkage disequilibrium levels did not significantly impair accuracy.
- Haplotype frequencies remained consistent even with the introduction of missing data.
Conclusions:
- Missing data, up to 30%, does not perceptibly affect the overall accuracy of haplotype analysis.
- Ambiguous data points should be treated as unknown in genetic analyses.
- Findings have significant implications for data collection strategies in genetic studies.