Biological Knowledge Integration in DNA Microarray Gene Expression Classification Based on Rough Set Theory

DNA microarrays have contributed to the exponential growth of genetic data from years. This large amount of gene expression data have been used in researches seeking diagnosis of diseases like cancer using classification methods. In turn, explicit biological knowledge about gene functions has also grown tremendously over the last decade. This work integrates explicit biological knowledge in classification process using Rough Set Theory, making it more effective. In addition, the proposed model is able to indicate which part of biological knowledge has been important for classification. The classification process is divided into five steps. Firstly, supergenes are created, which summarize the information of intersections of sets of genes (called basic categories)from biological knowledge using Principal Component Analysis. Then, continuous values of supergenes are discretized using Discriminant Fuzzy Patterns. The third step is to select the most relevant supergenes using the criterion of maximum β-relevance, supported by Rough Set Theory. Then, decision rules are generated using the CAI (Conjuntos Aproximados con Incertidumbre) model, which are the basis of the final classifier. Finally, a classifier is contructed using decision rules generated in the previous step, giving they an order of application based on a score. Based on a set of samples from DNA microarrays and explicit biological knowledge expressed as sets of genes that may or may not be related to the concept that seeks to be classified, the proposed model is evaluated, obtainin successful results compared to famous classification techniques.