MiRmat

MiRmat predicts mature microRNA sequences from primary microRNA transcripts (pri-miRNAs) by identifying Drosha and Dicer cleavage sites to map miRNA maturation and support downstream regulatory analysis.

Key Features:

• Mature microRNA prediction: Predicts mature miRNA sequences from pri-miRNAs by analyzing Drosha and Dicer cleavage sites.
• Drosha processing site prediction: Analyzes free energy distribution along the pri-miRNA secondary structure, using conserved patterns observed in vertebrate microRNA hairpins to identify Drosha cleavage sites.
• Dicer processing site prediction: Uses structural features of pre-miRNAs and a Random Forest classifier, considering free energy distribution downstream of the pri-miRNA secondary structure, to predict Dicer cleavage sites.
• Seeding sequence prediction: Determines the mature miRNA "seeding sequence" to infer target mRNA binding specificity.
• Genome-scale regulatory network inference: Facilitates mapping of genome-scale post-transcriptional regulatory networks based on predicted mature miRNAs.
• Performance: Identifies 77.8% of Drosha sites and 92.8% of Dicer sites within two nucleotides on an independent test set from ten vertebrates, and 71.9% (Drosha) and 87.2% (Dicer) under stringent family-exclusion conditions.

Scientific Applications:

• Novel MicroRNA Identification: Accurate prediction of mature miRNA sequences aids discovery and characterization of novel microRNAs across species.
• Target mRNA Binding Specificity: Predicts the seeding sequence of mature miRNAs to infer binding specificity to target mRNAs and analyze post-transcriptional regulatory interactions.
• Genome-Scale Regulatory Network Inference: Supports construction of genome-scale post-transcriptional regulation networks to study gene expression control mechanisms.

Methodology:

Analyzes free energy distribution patterns along pri-miRNA secondary structures to predict Drosha cleavage sites; uses structural features of pre-miRNAs and a Random Forest algorithm, considering downstream free energy distribution, to predict Dicer sites; and infers mature miRNA sequences from the predicted Drosha and Dicer cleavage positions, relying on conserved patterns observed in vertebrate microRNA hairpins.