MLIP: using multiple processors to compute the posterior probability of linkage

Area of Science:

• Genetics
• Computational Biology
• Bioinformatics

Background:

• Genetic linkage analysis for complex traits involves navigating vast multidimensional parameter spaces.
• The posterior probability of linkage (PPL) is a statistical method for human complex trait genetic mapping, designed for mathematical rigor.
• Calculating PPL is computationally intensive due to the need to evaluate complex integrals, hindering its application.

Purpose of the Study:

• To evaluate the effectiveness of a multiprocessor system (MLIP) in reducing computation time for PPL calculations.
• To assess the scalability and performance of MLIP in genetic linkage analysis.

Main Methods:

• Development and implementation of MLIP, a multiprocessor system for two-point genetic linkage analysis.
• Utilizing both simulated and real genetic data for performance evaluation.
• Analyzing PPL calculations across varying parameter spaces and data characteristics.

Main Results:

• MLIP significantly accelerates two-point log-likelihood ratio calculations over model parameter grids.
• Empirical results demonstrate substantial speedups in PPL computations.
• The system's performance is influenced by data characteristics like parameter grid granularity and pedigree structure.

Conclusions:

• The MLIP system enables efficient computation of the PPL, facilitating complex trait genetic mapping.
• Full multidimensional genome scans are now feasible within days, a significant improvement over previous runtimes.
• Ongoing optimization efforts aim to further enhance MLIP's performance.