An efficient parallel algorithm for haplotype inference based on rule based approach and consensus methods.

With recent advancements in bioinformatics technology, completion of ‘Human Genome Project’, and the exponential growth in genetic data, there is an ever-growing need in genomics to determine the relevance of genetic variations with observed phenotypes and complex diseases. The most common form of genetic variation is a single base mutation called SNP (Single Nucleotide Polymorphism), resulting in different alleles present at a given locus. Grouped together, a set of closely linked SNPs on a particular copy of chromosome is defined as a ‘Haplotype’. Research has proven that analysis of haplotypes is potentially more promising and insightful, and hence, at the forefront of bioinformatics investigations, especially due to its significance in the complex disease association studies. Current routine genotyping methods typically do not provide haplotype information, essential for many analyses of fine-scale molecular-genetics data. Biological methods for Haplotype Inference are cost prohibitive and labor intensive. Hence continues the search for more accurate computational methods for haplotype determination from abundantly and inexpensively available ambiguous genotype data. Continuing the search for a more efficient algorithm, we present in this work two parallel algorithmic approaches, based on ‘Inference Rule’ introduced by Clark in 1990 and the consensus method reported by Dr. Steven Hecht Orzack in 2003. One approach parallelizes the consensus method. As although, consensus method produces results comparable to other leading haplotype inference algorithms, its time efficiency can be significantly improved to further investigate this promising method for haplotype inference with larger data sets and greater number of iterations of Clark’s algorithm. The parallel algorithm is also used to study the affect of different number of iterations used for consensus method. Second parallel approach introduces an Enhanced Consensus algorithm that improves upon the average accuracy achieved by consensus method in much smaller time interval.