Evaluation of allele frequency estimation using pooled sequencing data simulation.

Area of Science:

• Genomics
• Bioinformatics

Background:

• Next-generation sequencing (NGS) provides detailed genomic analysis, particularly for disease association studies.
• NGS offers advantages over fixed SNP genotyping chips, with comparable costs for some applications.
• DNA pooling is a cost-saving strategy for large-scale genomic studies.

Purpose of the Study:

• To rigorously evaluate the accuracy of estimating allele frequencies from pooled next-generation sequencing data.
• To identify and quantify sources of error in pooled sequencing, including reference allele preference.

Main Methods:

• Development of a simulation model incorporating key factors: pool size, overall depth, average depth per sample, pooling variation, and sampling variation.
• Utilizing real DNA sequencing data to measure and implement reference allele bias within the simulation.

Main Results:

• Pooled sequencing data can significantly increase relative error rates in allele frequency estimation.
• Error rates are disproportionately higher for single nucleotide polymorphisms (SNPs) with low minor allele frequencies.
• Reference allele preference in sequencing data contributes to observed errors.

Conclusions:

• Pooled sequencing is practical for allele frequency estimation but requires careful consideration of potential errors.
• Recommendations for mitigating errors include employing large pool sizes and ensuring high average sequencing depth per sample.