Translation of Genome-wide expression into biological meaningful data is still challenging. Recent efforts have interpreted microarray data by using prior knowledge such as pathways. Even though pathway based gene expression analysis is widely used, all pathway level analysis methods utilize conventional pathways. However, it is not a proper to interpret microarray using conventional pathways, since conventional pathways consist of various interactions of signalling proteins and regulatory molecules. Therefore, we propose a new concept of pathways, Transcriptional Regulation Pathways (TRP), specifically design to capture the context specific transcriptional regulations for pathway level analysis in disease expression profiles. TRP consists of two parts, TF target genes and subset of pathway genes. Among TF target genes in database, we choose target genes whose expression correlates with TF by ARACNe which utilizes MI (Mutual Information). We select pathway member genes transcribed by TF in other pathways. Among those genes, we choose genes verified by ARACNe and Mapper algorithm which computationally searches for binding sites in promoter regions. We use Tian’s method for pathway level analysis. The results from Tian’s method provide that TRP shows better performance than normal pathways in the classification of Type 2 Diabetes and Lung cancer expression profiling. Our new proposed concept utilizes the trait of gene expression profiling so that it gives disease phenotype related pathways in pathway level analysis and good performance in cancer classification.