Markov chain Monte Carlo methods for radiation hybrid mapping

Area of Science:

• Genetics
• Computational Biology
• Statistical Genetics

Background:

• Genetic mapping relies on accurately ordering genetic loci.
• Markov chain Monte Carlo (MCMC) methods are used for genetic mapping, handling missing data and errors.
• Existing MCMC schemes for genetic mapping suffer from poor mixing due to parameter correlations.

Purpose of the Study:

• To investigate the impact of simulated tempering on MCMC mixing characteristics for genetic mapping.
• To assess the effectiveness of simulated tempering in radiation hybrid mapping.

Main Methods:

• The study employed a modified Markov chain Monte Carlo (MCMC) sampling scheme: simulated tempering.
• The method was applied to analyze haploid radiation hybrid mapping data.
• The performance was evaluated based on the mixing characteristics of the Markov chain.

Main Results:

• Simulated tempering significantly improved the mixing performance of the MCMC sampling scheme.
• The approach demonstrated enhanced efficiency in ordering genetic loci for radiation hybrid mapping.
• The effectiveness was noted for mapping smaller numbers of loci.

Conclusions:

• Simulated tempering is a valuable enhancement for MCMC-based genetic mapping, particularly for radiation hybrid mapping.
• While not ideal for very large numbers of loci (>100), it is suitable for fine-scale mapping of chromosomal subsections.