TargetProfiler

TargetProfiler predicts microRNA (miRNA) targets using a hidden Markov model trained on experimentally validated miRNA-target interactions and evaluated against protein downregulation datasets to improve biological correspondence.

Key Features:

• Probabilistic Learning Algorithm: Implements a hidden Markov model to capture patterns indicative of miRNA-target interactions.
• Training on Verified Data: Trained on a dataset of experimentally validated miRNA targets.
• Validation Against Protein Downregulation Data: Evaluated against large-scale protein downregulation datasets to assess correlation between predictions and protein-level effects.
• Comparative Performance: Compared with three other miRNA target prediction tools and demonstrated higher accuracy in predicting experimentally verified targets.

Scientific Applications:

• Identification of Novel miRNAs and Their Targets: Used to predict targets for a novel miRNA gene located in a cancer-associated genomic region.
• Experimental Validation: The predicted interaction between the novel miRNA and CCND2 was experimentally verified.

Methodology:

Uses a hidden Markov model trained on experimentally validated miRNA targets and evaluated against large-scale protein downregulation datasets, with comparative analyses versus three other miRNA target prediction tools.