Differentially expressed genes (DEGs) are commonly used to study the pathological mechanism of various brain diseases. However, these analyses are not sufficient to identify genes that alter their roles in the transcription networks but without significant changes in the expression level. By analyzing an RNA microarray dataset (GSE36980) from Gene Expression Omnibus (GEO), we find that many genes exhibit changes in how their transcriptions are correlated with that of the entire genome in the samples of Alzheimer’s disease (AD) patients, when compared with the normal controls. We name such genes as role-alteration genes (RAGs). Only a minority of RAGs also show significant changes in the expression level. Within the pools of RAGs, we further identified clustered biological functions through gene ontology (GO) analyses, suggesting the functional significance of these RAGs. Our work provides novel insights into the genes that may be involved in AD but cannot be captured by common DEG analyses, suggesting more profound changes in the transcription network during diseases than previously appreciated.