RRSM with a data-dependent threshold for miRNA target prediction

Wan J. Hsieh, Hsiuying Wang*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

4 Scopus citations


Predicting miRNA target genes is one of the important issues in bioinformatics. The correlation analysis is a widely used method for exploring miRNA targets through microarray data. However, the experimental results show that correlation analysis leads to large false positive or negative results. In addition, the correlation analysis is not appropriate when multiple miRNAs simultaneously regulate a gene. Recently, the relative R squared method (RRSM) has been proposed for miRNA target prediction, which is shown to be superior to some existing methods. To adopt the RRSM, we need first to set thresholds to select a proportion of potential targets. In the previous studies, the threshold is set to be fixed, which does not depend on the characteristic of a gene. Due to the diversity of the functions of genes, a data-dependent threshold may be more feasible in real data applications than a data-independent threshold. In this study, we propose a threshold selection method which is based on the distribution of the relative R squared statistic. The proposed method is shown to significantly improve the previous prediction results by selecting more experimentally validated targets.

Original languageEnglish
Pages (from-to)54-60
Number of pages7
JournalJournal of Theoretical Biology
StatePublished - 1 Nov 2013


  • Correlation analysis
  • P-value
  • Regression model
  • The relative R squared method


Dive into the research topics of 'RRSM with a data-dependent threshold for miRNA target prediction'. Together they form a unique fingerprint.

Cite this