A computational approach to the identification of target genes regulated by Systemic Sclerosis-associated miRNAs

Abstract

In the light of previous research (Chouri et al, unpublished results) performed on Systemic sclerosis (SSc), arises the importance of determining the molecular pathways affected by SScassociated miRNAs. To accomplish this, it is crucial to identify putative target genes affected by miRNAs that are dysregulated in SSc. The goal of our research is to implement a computational approach to identify putative target genes of miRNAs associated with SSc. We propose here an in-silico pipeline based on Pearson correlation of miRNAs and RNAseq data. We applied this method to a dataset from a cohort of SSc patients. Our approach also considers the multiple testing problem as well as different significance thresholds and various target resources. To evaluate the resulting interactions we determine the strength (significance) and type (negative correlation) of association and compare our approach with the method proposed by van Iterson et al. In the latter study (van Iterson et al, 2013), a multiple linear regression approach called the global test was proposed and was proved to successfully identify target genes by using an integrated analysis of miRNA and mRNA expression. It was also compared to correlation and LASSO methods in terms of predictive performance. We show that our method successfully identifies miRNA-mRNA interactions found in reliable target databases (published validated and experimentally supported) but can also identify potential novel interactions by investigating sequence-based predicted interactions. Finally, the most relevant findings of our analysis might be used for further validation through wet-lab experiments as a next step of investigation.