MIDClass is a simple and accurate method for gene expression classification. The method relies on associative classification and is based on the idea that the expression range of genes may efficiently identify subtypes in the same class. MIDClass can be summarized as follows. First, a statistical test is applied to filter out the uniformative genes whose expression do not not present any significant change across the classes. Next, a discretization algorithm allows to partition the gene expression range into subranges presenting strong discriminant power in each class. In the last phase the gene expression ranges are treated as item. Each phenotype of a class is considered as a record containing a set of items. Therefore, phenotypes of each class are given as input to an algorithm to extract maximal frequent itemset which will characterize subclasses. Those frequent item sets are then used as rules in which the antecedent part is the combination of genes expression interval, and the consequence is the class-label.
Before downloading, be sure to have Java installed. The sources are available upon request, but for inclusion in other applications, we recommend to use the JAR file as a JAVA library, or the CLI version of the application.
For bugs reporting, please send an email to: firstname.lastname@example.org