miRNA-dis

miRNA-dis applies Support Vector Machines to predict microRNA precursor (pre-miRNA) sequences using distance structure status pair and distance-pair feature representations to support identification of pre-miRNAs for studies of gene regulation.

Key Features:

• Feature Representation: Constructs feature vectors based on the "distance structure status pair" and "distance-pair" to capture sequence-order and structural characteristics of pre-miRNAs.
• Classifier: Uses Support Vector Machines (SVM) as the predictive model.
• Validation: Performance was assessed by rigorous cross-validation on a newly constructed benchmark dataset.
• Predictive Accuracy: Reports a prediction accuracy of 87.02% for human pre-miRNAs and demonstrates performance across animals, plants, and viruses.
• Interpretable Model: Provides model interpretability enabling analysis of discriminative sequence features characteristic of microRNAs.
• High Throughput Capability: Supports high-throughput analysis of microRNA precursor candidates.

Scientific Applications:

• Gene Regulation Studies: Identification of pre-miRNAs to facilitate investigations into microRNA-mediated gene regulatory mechanisms.
• Comparative Genomics: Prediction of pre-miRNAs across animals, plants, and viruses to support cross-species comparative and evolutionary analyses.
• Disease Research: Analysis of microRNA precursor dynamics relevant to disease mechanisms such as cancer and viral infections.

Methodology:

Constructs feature vectors using the "distance structure status pair" and "distance-pair", trains Support Vector Machines (SVM), and evaluates performance via rigorous cross-validation on a newly constructed benchmark dataset.