Skip to Main Content
MicroRNAs (miRNAs) are reported to play essential roles in cancer initiation and progression and microarray technologies are intensively applied to study the miRNA expression profile in cancer. It is very common that the set of differentially expressed miRNAs related to the same cancer identified from different laboratories varies widely. Meanwhile, how the altered miRNAs coordinately contribute to the cause of prostate cancer is still not clear. In this study, we collected and processed four human prostate cancer associated miRNA microarray expression datasets with newly developed cancer outlier detection methods to identify differentially expressed miRNAs (DE-miRNAs). The targets of these DE-miRNAs were then extracted from database or predicted by bioinformatics prediction and then mapped to functional databases for enrichment analysis and overlapping comparison. Newly developed outlier detection methods were found to be more appropriate than t-test in cancer research, and the consistency of independent prostate cancer expression profiles at pathway or gene-set level was shown higher than that at gene (i.e. miRNA here) level. Furthermore, we identified 41 Gene Ontology terms, 4 KEGG pathways and 77 GeneGO pathways which are associated with prostate cancer. Among the top 15 GeneGO pathways, 5 were reported previously and the rest could be putative ones. Our analyses showed that more appropriate outlier detection methods should be used to detect oncogenes or oncomiRNAs that are altered only in a subset of samples. We proved that expression signatures of independent microarray experiments are more consistent rather at pathway level than at miRNA / gene level. We also found that the utilization of similar meta-analysis methods between miRNA and mRNA profiling datasets result in the detection of the same pathways.