Estimating haplotype frequencies in pooled DNA samples when there is genotyping error

Area of Science:

• Population Genetics
• Bioinformatics

Background:

• Haplotype frequencies are crucial for genetic studies.
• The expectation maximization (EM) algorithm is used for estimating haplotype frequencies from pooled DNA.
• Genotyping errors can potentially affect the accuracy of these estimates.

Purpose of the Study:

• To evaluate the impact of genotyping error on haplotype frequency estimation using the EM algorithm.
• To investigate how factors like allele frequency, linkage disequilibrium (LD), genotyping error rate, and pool size influence estimation accuracy.

Main Methods:

• Simulations were conducted to model haplotype frequency estimation.
• The study explored various model parameters including allele frequencies, inter-marker LD, genotyping error rates, and pool sizes (2, 5, and 10 individuals).

Main Results:

• Pool sizes of 2, 5, and 10 individuals demonstrated comparable accuracy in haplotype frequency estimation.
• Less accurate estimates were observed with common marker allele frequencies and in the absence of inter-marker LD, irrespective of genotyping error.
• Genotyping error showed a slight decrease in the accuracy of haplotype frequency estimates.

Conclusions:

• The EM algorithm remains robust and performs well even with the presence of genotyping errors.
• Small pool sizes (2-10 individuals) provide similar accuracy for haplotype frequency estimates while significantly reducing genotyping costs.
• These findings support the use of pooled DNA and the EM algorithm for cost-effective haplotype frequency estimation.