Email updates

Keep up to date with the latest news and content from BMC Bioinformatics and BioMed Central.

This article is part of the supplement: Selected papers from the Seventh Asia-Pacific Bioinformatics Conference (APBC 2009)

Open Access Research

A fast algorithm for genome-wide haplotype pattern mining

Søren Besenbacher12*, Christian NS Pedersen12 and Thomas Mailund1

Author Affiliations

1 Bioinformatics Research Center, University of Aarhus, Denmark

2 Department of Computer Science, University of Aarhus, Denmark

For all author emails, please log on.

BMC Bioinformatics 2009, 10(Suppl 1):S74  doi:10.1186/1471-2105-10-S1-S74

Published: 30 January 2009



Identifying the genetic components of common diseases has long been an important area of research. Recently, genotyping technology has reached the level where it is cost effective to genotype single nucleotide polymorphism (SNP) markers covering the entire genome, in thousands of individuals, and analyse such data for markers associated with a diseases. The statistical power to detect association, however, is limited when markers are analysed one at a time. This can be alleviated by considering multiple markers simultaneously. The Haplotype Pattern Mining (HPM) method is a machine learning approach to do exactly this.


We present a new, faster algorithm for the HPM method. The new approach use patterns of haplotype diversity in the genome: locally in the genome, the number of observed haplotypes is much smaller than the total number of possible haplotypes. We show that the new approach speeds up the HPM method with a factor of 2 on a genome-wide dataset with 5009 individuals typed in 491208 markers using default parameters and more if the pattern length is increased.


The new algorithm speeds up the HPM method and we show that it is feasible to apply HPM to whole genome association mapping with thousands of individuals and hundreds of thousands of markers.