We revealed the feature pathways by computing the classification error rates of out-of-bag (OOB) by random forests combined with pathway analysis. At each feature pathway, the relativity of gene expression was studied and the co-regulated gene patterns under different experiment conditions were analyzed by MAP (Mining attribute profile) algorithm. The discovered patterns were also clustered by the average-linkage hierarchical clustering technique. The results showed that the expression of genes at the same pathway was similar. The co-regulated patterns were found in two feature pathways of which one contained 108 patterns and the other contained 1 pattern. The results of clusters showed that the smallest Pearson coefficient of the clusters was more than 0.623, indicating that the co-regulated patterns in different experiment conditions were more similar at the same KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway. The methods can provide biological insight into the study of microarray data.