https://www.nature.com/articles/nsmb.2678
We extracted miRNA-mRNA pairs with strong negative REC score (REC < −6.2, FDR < 2 × 10−4, 4,584 pairs with less than one estimated false positive) and evidence for target interaction as predicted by miRanda (score < −0.5), TargetScan (context score < −0.2) and evolutionary conservation (TargetScan probability of conserved targeting, PCT, >0.5). These thresholds were chosen to obtain a high-confidence list of candidate miRNA-target interactions with possible functional roles across a range of cancer types. The combination of the REC score and target prediction filters yielded 143 miRNA-mRNA pairs (Fig. 4a), significantly more than was expected by chance (P = 3.1 × 10−85, two-tailed binomial test, k = 143, n = 4,584, r = 3.4 × 10−3 = 22,589 predicted targets / 6,642,349 total pairs), consistent with the hypothesis that the REC score can be used to augment sequence-based miRNA-target predictions and infer functionally relevant target interactions in vivo. These 143 putative recurring target interactions formed a network of 40 evolutionarily conserved miRNAs and 72 target mRNAs (Fig. 4b and Supplementary Table 1).
Tools for Sequence-Based miRNA Target Prediction: What to Choose?
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5187787/
RNA-Seq differential expression work flow using DESeq2
http://www.sthda.com/english/wiki/print.php?id=55
Precision Medicine Bioinformatics
Introduction to bioinformatics for DNA and RNA sequence analysis
RNAseq Analysis
https://pmbio.org/module-06-rnaseq/0006/01/01/Intro_to_RNAseq_Analysis/