Haplotype inference based on Hidden Markov Models in the QTL-MAS 2010 multi-generational dataset

Area of Science:

• Genetics
• Computational Biology
• Bioinformatics

Background:

• Previous work demonstrated efficient computation of genotype probabilities and allele inheritance.
• An extension for haplotype inference (phasing) using Hidden Markov Models was developed.
• Computational phasing of large marker datasets remains a challenge.

Purpose of the Study:

• To investigate an improved method for haplotype inference in multi-generational datasets.
• To apply and evaluate the method for quantitative trait loci (QTL) detection.

Main Methods:

• Utilized Hidden Markov Models for haplotype inference.
• Adapted Markov model training for convergence in complex datasets.
• Applied the method to a simulated multi-generational dataset for QTL detection.

Main Results:

• Achieved zero deviations in phasing for the founder generation compared to simulated data.
• Successfully phased 99.93% of all markers.
• Ensured 97.68% of individuals were correctly phased across all markers and chromosomes.
• Results were obtained within a weekend on a small computational cluster.

Conclusions:

• The developed method offers efficient and near-perfect haplotype inference for dense pedigrees.
• Enables complete genome phasing, valuable for tracking allele origins and complex crosses.
• The cnF2freq codebase is available under a BSD-style license.