Enhancing miRNA annotation confidence in miRBase by continuous cross dataset analysis

The immaculate annotation of all microRNAs (miRNAs) is a prerequisite to study their biological function on a genome-wide scale. However, the original criteria for proper miRNA annotation seem unsuited for the automated analysis of the immense number of small RNA reads available in next generation sequencing (NGS) datasets. Here we analyze the confidence of past miRNA annotation in miRBase by cross-analyzing publicly available NGS datasets using strengthened annotation requirements. Our analysis highlights that a large number of annotated human miRNAs in miRBase seems to require more experimental validation to be confidently annotated. Notably, our dataset analysis also identified almost 300 currently non-annotated miRNA*s and 28 novel miRNAs. These observations hereby greatly increase the confidence of past miRNA annotation in miRBase but also illustrate the usefulness of continuous re-evaluating NGS datasets in the identification of novel miRNAs.