Jianwei Li*, Leibo Liu, Qinghua Cui and Yuan Zhou* Pages 1104 - 1112 ( 9 )
Background: De-regulation of microRNAs (miRNAs) is closely related to many complex diseases, including cancers. In The Cancer Genome Atlas (TCGA), hundreds of differentially expressed miRNAs are stored for each type of cancer, which are hard to be intuitively interpreted. To date, several miRNA set enrichment tools have been tailored to predict the potential disease associations and functions of de-regulated miRNAs, including the miRNA Enrichment Analysis and Annotation tool (miEAA) and Tool for Annotations of human MiRNAs (TAM1.0 &TAM 2.0). However, independent benchmarking of these tools is warranted to assess their effectiveness and robustness, and the relationship between enrichment analysis results and the prognosis significance of cancers.
Methods: Based on differentially expressed miRNAs from expression profiles in TCGA, we performed a series of tests and a comprehensive comparison of the enrichment analysis results of miEAA, TAM 1.0 and TAM 2.0. The work focused on the performance of the three tools, disease similarity based on miRNA-disease associations from the enrichment analysis results, the relationship between the overrepresented miRNAs from enrichment analysis results and the prognosis significance of cancers.
Results: The main results show that TAM 2.0 is more likely to identify the regulatory disease’s functions of de-regulated miRNA; it is feasible to calculate disease similarity based on enrichment analysis results of TAM 2.0; and there is weak positive correlation between the occurrence frequency of miRNAs in the TAM 2.0 enrichment analysis results and the prognosis significance of the cancer miRNAs.
Conclusion: Our comparison results not only provide a reference for biomedical researchers to choose appropriate miRNA set enrichment analysis tools to achieve their purpose but also demonstrate that the degree of overrepresentation of miRNAs could be a supplementary indicator of the disease similarity and the prognostic effect of cancer miRNAs.
miRNAs, miRNA set enrichment analysis, miRNA-disease association, cancer prognosis, disease similarity, genome.
Institute of Computational Medicine, School of Artificial Intelligence, Hebei University of Technology, Tianjin 300401, Institute of Computational Medicine, School of Artificial Intelligence, Hebei University of Technology, Tianjin 300401, Department of Biomedical Informatics, School of Basic Medical Sciences, Center for Noncoding RNA Medicine, Peking University, Beijing 100191, Department of Biomedical Informatics, School of Basic Medical Sciences, Center for Noncoding RNA Medicine, Peking University, Beijing 100191