Institute for Biomedical Technologies - National Research Council (ITB-CNR), Segrate, Italy.
Brief Bioinform. 2011 Nov;12(6):588-600. doi: 10.1093/bib/bbr062. Epub 2011 Oct 22.
miRNA target genes prediction represents a crucial step in miRNAs functional characterization. In this context, the challenging issue remains predictions accuracy and recognition of false positive results. In this article myMIR, a web based system for increasing reliability of miRNAs predicted targets lists, is presented. myMIR implements an integrated pipeline for computing ranked miRNA::target lists and provides annotations for narrowing them down. The system relies on knowledge base data, suitably integrated in order to extend the functional characterization of targeted genes to miRNAs, by highlighting the search on over-represented annotation terms. Validation results show a dramatic reduction in the quantity of predictions and an increase in the sensitivity, when compared to other methods. This improves the predictions accuracy and allows the formulation of novel hypotheses on miRNAs functional involvement.
miRNA 靶基因预测是 miRNA 功能表征的关键步骤。在这方面，仍然存在预测准确性和识别假阳性结果的难题。本文介绍了 myMIR，这是一个基于网络的系统，用于提高 miRNA 预测靶标列表的可靠性。myMIR 实现了一个用于计算排名 miRNA::target 列表的集成管道，并提供注释来缩小它们的范围。该系统依赖于知识库数据，通过突出显示对过度表示的注释术语的搜索，适当集成以将靶向基因的功能表征扩展到 miRNA。验证结果表明，与其他方法相比，预测的数量显著减少，而灵敏度增加。这提高了预测的准确性，并允许对 miRNA 功能参与提出新的假设。
