Meta-analysis of liver transcriptomic data identifies accurate disease classifiers and disease perturbed networks

microarray;meta-analysis;disease transcriptomic signature;network perturbation

Wang, Yuliang ; Price ; Nathan D.

关键词: microarray;    meta-analysis;    disease transcriptomic signature;    network perturbation;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/24023/Wang_Yuliang.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Chronic liver diseases are a major health problem. Previous DNA microarray studies of different liver diseases have improved our knowledge of the molecular pathogenesis of liver diseases and produced potential biomarkers. However, these studies typically rely on binary phenotype comparisons (e.g. cancer vs. normal) to identify disease signatures. It is possible that the resulting signatures may be partially shared by other liver diseases not included in the binary comparison. In this study, we took a comprehensive and organ-specific approach, where we studied all liver pathophysiological states in a single unifying context, and found a specific transcriptomic signature for each phenotype with respect to all the other phenotypes, instead of just one. The resulting 36-gene disease signature had 85% accuracy in 10 fold cross validation. Through stringent leave-one-lab out independent validation, we found that high classification accuracy was achieved when there was a total of around 100 samples from 2 independent contributing labs. We also identified perturbed networks in liver diseases in general and hepatocellular carcinoma in particular. Many of the classifier genes and perturbed networks are involved in important biological processes in liver disease pathogenesis, including immune response and inflammation, fibrogenesis, metabolism and its regulation, apoptosis, and cellular signaling. The disease classifiers and perturbed networks identified in this study may be potential candidates for novel diagnostic approaches to multiple liver diseases.