Identification of microRNA target sites: in-silico screening on NGS data

MiRNAs may be biomarkers associated with the progression of several diseases and great eff orts have been made to discover miRNAs, identify miRNA targets and infer miRNA functions. Diff erent bioinformatic tools and algorithms enabled to predict miRNAs binding regions together with several databases that provided information about miRNA sequences. WGS, WES and RNAseq datasets can represent an optimal support to correlate variants with miRNA target sites at no additional cost. Our goal is to link variant calling information with a potential gain or loss of miRNAs target sites. So far, there is no simply logical and sequential procedure to do this. Using already existing software, we developed a simple pipeline that allows us to identify the impact of variants on miRNA target sites with a biological role. We created a unique database of miRNA's sequences that collected all existing miRNA databases (miRBase, tarbase, mirRSNP, etc) and performed prediction analyses of target sites focused on the most performing existing algorithms (MiRanda, TargetScan). For example, starting from about 23,000 genomic variants that lie in 3'UTR region, a simple input command enabled us to select 4637 lost and 4302 gained target sites. The output is a simple tabular text fi le contains genomic variants, target site, variant impact, associated miRNAs, loss and gain sites and annotation data coming from diff erent databases. The output result was designed to be used in subsequent standard software to apply fi lters based on the desired mode of inheritance and on sample’s genotype quality and variant frequencies. This method could be particularly useful on WES data when the identifi cation of the causal variant has failed and the involvement of miRNAs is assumed, without additional costs. This new automated sequential pipeline allows to identify, in silico, the variants involved in the miRNA targets from VCF fi les and to generate results easily interpreted and usable by people without bioinformatic skills.