Frequent itemset mining for discovering co-occurring genomic tracks

Web proceedings papers

Authors

Boris Simovski and Ana Madevska-Bogdanova

Abstract

The problem of finding subsets of genomic tracks that occur at same genomic locations is a complicated and resource exhaustive task. To address the issue we turn to frequent itemset mining and the well known algorithms Apriori and Eclat. Because of the genomic tracks’ different densities, the results are on- ly partial. As a solution we propose a new algorithm, the HBExpectedSupportMiner. This algorithm uses Apriori as the basis for its search space exploration strategy. But, in order to determine if an itemset is frequent, our algorithm uses the expected support of the itemset. The expected support of an itemset is dependent on the density of its items. Hence the results give better insight into the patterns in the data.

Keywords

Frequent itemset mining, Genomic track, Genomic HyperBrowser, Apriori, Eclat