Unveiling Fuzzy Associations Between Breast Cancer Prognostic Factors and Gene Expression Data

Breast cancer is the second most common cancer worldwide and the fifth most common cause of cancer death. There are many prognostic factors associated with breast cancer which are usually considered when determining how cancer will affect a patient. In addition, distinct molecular subtypes of breast tumors have been described by gene expression profiling. In this work we integrate information from the main prognostic factors in breast cancer with whole-genome microarray data to study the potential associations between these two types of data. The heterogeneity and noisy nature of the data along with its high dimensionality make necessary the use of data mining techniques to analyze the dataset. Fuzzy sets are particularly suitable to model imprecise and noisy data, while association rules are very appropriate to deal with heterogeneous and high dimensionality data. Thus, a fuzzy association rule mining algorithm was used to carry out this study. Many interesting associations have been obtained. Further studies and empirical evaluation of these associations are needed to obtain scientific evidence of such relations. Finally, a freely accessible web application has been developed which implements the fuzzy association rule mining algorithm used in this study (http://genome.ugr.es/biofar).