Ranking Analysis of Gene Expression and Methylation Data for Identification of Weighted Association Rules
Manju Priya,  Durai Kumar.D,  Balaji.S
Association rule mining is an interesting topic in data mining and bioinformatics. The huge number of evolved rules by association rule mining algorithms makes confusion to the decision maker. In this paper, three techniques for mining association rules are proposed. For selection of perfectly Differentially Expressed (DE)/ Methylated(DM) genes the p-value and fold change value are used. For assigning weight to each gene the weighted condensed support (wcs) and confidence (wcc) is used along with various network weighting methods which reduces the complexity. For data discretization effective cluster algorithm i.e., Genetic Cluster Algorithm (GCLUS) is used. Thus, it saves time for execution of algorithm. The genes of the top rules are biologically validated by Gene Ontology (GOs) and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analyses. Many top ranked rules extracted using these techniques hold poor rank in traditional Apriori algorithm which is significant to treat many diseases.
Keywords- rule mining, p-value, fold change value, wcs, wcc, network weighting methods, GCLUS
Cite this Article
Manju Priya,  Durai Kumar.D,  Balaji.S,   "Ranking Analysis of Gene Expression and Methylation Data for Identification of Weighted Association Rules"
, International Journal of Engineering Development and Research (IJEDR), ISSN:2321-9939, Volume.4, Issue 2, pp.1190-1192, May 2016, Available at :http://www.ijedr.org/papers/IJEDR1602209.pdf