HaploCart: Human mtDNA haplogroup classification using a pangenomic reference graph.

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Current mitochondrial DNA (mtDNA) haplogroup classification relies on single reference genomes, introducing bias and limiting uncertainty assessment.
• Existing methods struggle with low-coverage or incomplete sequence data, affecting assignment accuracy.

Purpose of the Study:

• To introduce HaploCart, a novel probabilistic mtDNA haplogroup classifier utilizing a pangenomic reference graph and Bayesian inference.
• To demonstrate HaploCart's superior performance compared to existing tools, particularly with challenging sequence data.
• To provide phylogenetically-aware, unbiased confidence scores for haplogroup assignments.

Main Methods:

• Developed HaploCart, a probabilistic classifier employing a pangenomic reference graph framework.
• Integrated principles of Bayesian inference for robust haplogroup assignment.
• Designed HaploCart as a command-line tool and web interface accepting FASTA, FASTQ, or GAM files.

Main Results:

• HaploCart significantly outperforms existing mtDNA haplogroup classification tools.
• The tool demonstrates robustness with low-coverage and incomplete consensus sequences.
• HaploCart generates phylogenetically-aware confidence scores, unbiased towards specific haplogroups.
• Reduced data requirements for confident haplogroup assignment were achieved.

Conclusions:

• HaploCart provides a more accurate and reliable method for mtDNA haplogroup classification.
• The pangenomic reference graph and Bayesian inference approach enhance classification robustness and confidence.
• HaploCart offers a valuable, user-friendly tool for genetic research, improving efficiency and data requirements.