Avoiding misleading estimates using mtDNA heteroplasmy statistics to study bottleneck size and selection.

Area of Science:

• Genetics
• Molecular Biology
• Evolutionary Biology

Background:

• Mitochondrial DNA (mtDNA) heteroplasmy is crucial for understanding genetic and developmental processes.
• Current methods using sample statistics for mtDNA selection and bottleneck size estimation are flawed.
• These flaws can lead to inaccurate conclusions, including false positive signals of selection.

Purpose of the Study:

• To address limitations in current statistical methods for analyzing mitochondrial DNA heteroplasmy.
• To introduce a more robust computational approach for analyzing mtDNA data.
• To improve the accuracy of hypothesis testing in mitochondrial genetics.

Main Methods:

• Developed a maximum likelihood (ML) approach for analyzing heteroplasmy data.
• Applied ML fits and explicit hypothesis tests for data analysis.
• Re-analyzed published experimental datasets using the new ML method.

Main Results:

• Identified that sample statistics often yield suboptimal fits to the Kimura distribution.
• Demonstrated that sample variance can be misleading for bottleneck size estimation, especially in small samples.
• Found that previously reported departures from the Kimura model in published datasets were not supported by re-analysis.

Conclusions:

• The proposed maximum likelihood approach is computationally simple and resolves issues with current methods.
• Advocates for the use of ML fits and explicit hypothesis tests over summary statistics and Kolmogorov-Smirnov tests.
• This new method provides a more reliable framework for studying mitochondrial DNA heteroplasmy mechanisms.