ViralMir

ViralMir identifies viral microRNA (miRNA) precursors using sequence and secondary-structure features to predict pre-miRNAs in viruses.

Key Features:

• Prediction target: Detects viral microRNA (pre-miRNA) precursors.
• Input data: Uses sequencing fragments derived from viral and human genomes as positive and negative examples.
• Training dataset: Trained on 263 experimentally validated pre-miRNAs from 26 virus species.
• Feature set: Employs 54 features extracted from RNA sequences and their secondary structural information.
• Algorithms: Implements support vector machines (SVMs) and random forest models for classification.
• Performance: Achieves a balanced accuracy exceeding 83% and outperforms previously developed pre-miRNA prediction tools.

Scientific Applications:

• Virus–host interaction analysis: Identification of viral miRNA precursors to study miRNA-mediated modulation of host gene expression.
• Viral pathogenesis research: Support investigation of viral roles in infection and host responses via predicted viral miRNAs.
• Posttranscriptional regulation studies: Facilitate research into miRNA effects on processes such as development, oncogenesis, and apoptosis.

Methodology:

Models (support vector machines and random forest) were trained on a dataset of 263 experimentally validated viral pre-miRNAs from 26 virus species and negative sequencing fragments from viral and human genomes using 54 sequence- and secondary-structure-derived features, achieving >83% balanced accuracy.