Classification of gene expression levels using activator and repressor motifs

Gene expression levels are influenced significantly by the presence or absence of cis-regulatory elements or motifs. This paper presents classification systems in which the occurrences of both activator and repressor motifs constitute important inputs in predicting whether a gene will be up-regulated, down-regulated, or neither (neutral). We have experimented with several approaches for classification using these input data, and best performance was obtained using Support Vector Machine (SVM) models with linear kernels and a hierarchical structure. On Saccharomyces cerevisiae data, this approach yielded 71% accuracy (on test data) for 3-category classification.