Figure below demonstrates the pipeline of miRTar2GO. In data processing step, genomic coordinates of the CLIP-Seq reads in different cell lines are mapped to the mRNAs to identify 3' UTRs which are enriched with AGO interaction sites.
In miRNA-mRNA allocation step, the 6mer seed region of all known miRNAs are aligned to the reverse complementary sequence of AGO CLIPed sites in the 3' UTRs. The result of this step is a set of miRNA-mRNA candidates where the miRNA seed region of each candidate has a perfect binding site on its associated mRNA.
In prediction categorization step, the whole miRNA sequence of each miRNA-mRNA is folded upon it's allocated CLIPed sequence to calculate the hybridization energy. At this step, interactions with a MFE greater than -15 are discard. The miRNA-mRNA interactions which are not known to be experimentally verified are selected to generate the unseen set. The miRNA-mRNA interactions which are known to be experimentally verified are used to calculated energy values for each miRNA including the minimum and the maximum of MFEs.
In result population step, four different set of result sets are generated. Those interactions in the unseen set with a MFE less than -20 introduce the result of miRTar2GO sensitive. Interaction scores are calculated for all pairs of the unseen introducing the highly sensitive miRTar2GO result. Those interactions of the unseen set which have a hybridization energy of less than -20 and less than the maximum MFE value associated to the miRNA are introduced as miRTar2GO highly specific results. Those interactions of the unseen set which have a hybridization energy of less than the maximum energy associated to their corresponding miRNA are introduced as miRTar2GO specific results.
Please note that miRTar2GO does not introduce the experimentally validated interactions in its result set as that data is used to train the prediction model.