4.4. In Silico Target Prediction Analysis

Two bioinformatics software packages, DIANA microT-CDS v. 5.0 (http://diana.imis.athena-innovation.gr/DianaTools/index.php?r=microT_CDS/index, Thessaly, Greece; accessed on 15 September 2020) [37,67] and TargetScanFish v. 6.2 (http://www.targetscan.org/fish_62/, Cambridge, MA, USA; accessed on 15 September 2020) [38] were programmed to predict the putative miRNA:mRNA interaction sites applying: the miRNA sequence; the 3′-UTR sequence; the Watson–Crick base pairing in 5′ end of the miRNA, called seed region; the free energy expressed in kcal/mol, a measure of binding stability; the 3′ region of the miRNA that also have a Watson–Crick base pairing with the mRNA; the level of conservation of this interaction between species; other regions with Watson–Crick base pairing in the 5′-UTR, open read frame (ORF) and CDS; and other factors unrelated to the Watson–Crick base pairing that can affect the miRNA action, called context. Only predictions with a miTG score > 0.7 and score + context < −0.2, respectively, were considered as effective miRNA targets.

The two prediction lists were overlapped, and a set of mutually predicted targets was generated for all miRNAs. The predicted genes that were retired from the ensemble were deducted from each software prediction list.