Identifying the largest complete data set from ALFRED
Mohamed Uduman · RIT Scholar Works (Rochester Institute of Technology) · 2006
ALFRED is a central and curated repository for allele frequency data for anthropologically defined human populations. To study and estimate the relationships and similarities between populations, researchers require a large and complete data set. However, the data set within ALFRED is not complete. Specifically, not all the populations in the database have been typed for all the polymorphisms. Mining ALFRED for the largest complete data set is equivalent to the 'Maximal Biclique' problem in graph theory. This is proven to be NP-Complete and no single algorithm can find the perfect solution in polynomial time. This project describes a heuristic (Largest Maximal Biclique Heuristic) which finds the largest complete data set from ALFRED, in real time. The program is compared to various other methods, including Wen- Chieh Chang's implementation of the 'maximal biclique' algorithm proposed by Alexe et.al. The algorithm efficiently mines ALFRED to extract the largest complete data set, and the results are made available for researchers in uniform data exchange format, through a Web site. Since ALFRED is updated frequently, the LMBH program is set up to mine ALFRED on a regular basis and provide researchers with the most up-to-date, largest complete data set from ALFRED.