TarP: A microRNA target gene prediction tool utilizing a polymorphic structured alignment approach.

MicroRNAs (miRNAs) represent a vital class of small non-coding RNAs that play key regulatory roles in gene expression. Accurate identification of miRNA-mRNA interactions is essential for understanding their biological functions. However, current computational prediction tools suffer from several limitations, including species-specific biases, suboptimal accuracy, high false discovery rates, and incomplete target gene coverage. To address these challenges, we present TarP, a novel miRNA target prediction algorithm employing a Polymorphic structured alignment (PMS) approach. Our method mimics the natural binding process between miRNAs and their target mRNAs by integrating key biological interaction features. The algorithm utilizes five distinct nucleotide-binding motifs to perform a structured decomposition and alignment of potential mRNA targets. Predictions are then rigorously evaluated through a dual scoring system: a Structure (St) coefficient assessing binding conformation and an Energy (En) coefficient evaluating thermodynamic stability, ensuring high-confidence target selection. Using experimentally validated human miRNA-mRNA interaction datasets, we benchmarked TarP against four widely used prediction tools (miRanda, RNAhybrid, PITA, and TargetScan). Comparative analyses demonstrate that TarP achieves superior performance in both sensitivity and specificity, exhibiting enhanced accuracy in positive target identification and improved discrimination between true and false interactions. The TarP algorithm is freely available at: https://github.com/Whimonk/TarP.