Inferring ethnicity from mitochondrial DNA sequence.

Area of Science:

• Genetics
• Forensic Science
• Bioinformatics

Background:

• Inferring coarse ethnic groupings from DNA is crucial for forensic investigations.
• Mitochondrial DNA (mtDNA) is a valuable genetic marker due to its maternal inheritance, high copy number, and persistence in degraded samples.
• The hypervariable region of the mitochondrial genome offers potential for ethnicity inference.

Purpose of the Study:

• To compare the performance of different methods for inferring ethnicity from mitochondrial DNA sequences.
• To evaluate the accuracy of support vector machines (SVM) for ethnicity inference.
• To identify the most informative segments of the mitochondrial hypervariable region and assess data handling strategies.

Main Methods:

• Utilized datasets from the mtDNA population database for comprehensive experiments.
• Applied support vector machines (SVM) for ethnicity inference.
• Compared SVM performance against nearest neighbor and discriminant analysis methods.
• Evaluated strategies for handling missing data in DNA sequences.

Main Results:

• Support vector machines (SVM) achieved an overall accuracy of 80-90% for ethnicity inference.
• SVM consistently outperformed previously proposed nearest neighbor and discriminant analysis methods.
• Identified the most informative segments within the mitochondrial hypervariable region.
• Determined that using common sequence regions is the optimal strategy for handling missing data.

Conclusions:

• Support vector machines (SVM) can accurately infer coarse ethnicity from limited mitochondrial DNA sequence data.
• The SVM approach demonstrates high accuracy, making it suitable for forensic applications.
• The findings suggest SVM algorithms are broadly applicable to other DNA sequence classification tasks.
• Optimal handling of missing data involves utilizing only common sequence regions between training and test data.