A Continuous Statistical Phasing Framework for the Analysis of Forensic Mitochondrial DNA Mixtures.

Area of Science:

• Genetics
• Bioinformatics
• Forensic Science

Background:

• Massively parallel sequencing generates valuable quantitative data.
• Resolving mitochondrial DNA (mtDNA) mitotypes from mixtures presents significant challenges, especially at varying ratios.

Purpose of the Study:

• To develop and validate a novel bioinformatic mixture deconvolution method for mtDNA analysis.
• To automate the reconstruction of complete individual mitotypes from mixed samples.

Main Methods:

• A population-based phasing approach was developed for mixture deconvolution.
• The method was tested using extensive in silico and in vitro two-person mixtures.
• Reference panels with known haplotypic variation were utilized to optimize phasing accuracy.

Main Results:

• Deconvolution accuracy was highest when mixture haplotypes matched reference panel haplotypes.
• Optimal accuracy was observed at mixture ratios that were neither extremely imbalanced nor nearly equal (e.g., 4:1).
• Haplotype estimation errors were primarily influenced by the accuracy of the input genotype data.

Conclusions:

• The developed framework is the first automated approach for reconstructing individual mitotypes from mixtures.
• This method successfully addresses challenging mixture ratios previously considered problematic in mtDNA analysis.