Fast computation of genetic likelihoods on human pedigree data

Area of Science:

• Computational biology
• Genetic epidemiology
• Bioinformatics

Background:

• Gene mapping and genetic epidemiology rely on computationally intensive likelihood calculations from human pedigree data.
• Complex pedigrees, numerous loci, and missing data can significantly impede calculation speed.
• Existing computational methods face challenges in efficiently processing large-scale genetic datasets.

Purpose of the Study:

• To introduce a novel array factorization method for accelerating genetic linkage calculations.
• To explore the application of vectorization and parallelization techniques to enhance computational speed in genetic analysis.
• To demonstrate the effectiveness of these methods using the MENDEL program on complex genetic problems.

Main Methods:

• Development and application of a new array factorization technique for linkage analysis.
• Implementation of vectorization for MENDEL on a supercomputer (IBM 3090).
• Implementation of parallelization for MENDEL on diverse parallel architectures and workstation networks.

Main Results:

• The array factorization method substantially accelerates linkage calculations with large marker sets.
• Vectorized and parallelized versions of MENDEL demonstrated significant improvements in computational speed.
• These optimized methods successfully addressed challenging linkage problems in human genetics.

Conclusions:

• The novel array factorization method, combined with hardware acceleration techniques like vectorization and parallelization, offers substantial speed improvements for genetic linkage analysis.
• These advancements are crucial for advancing gene mapping and genetic epidemiology, particularly with complex datasets.
• The revised MENDEL program provides a more efficient tool for large-scale genetic computations.