MiRDeep

MiRDeep identifies microRNAs (miRNAs) from deep sequencing data by modeling miRNA biogenesis to distinguish genuine miRNAs from other small RNAs.

Key Features:

• Probabilistic modeling: Employs a probabilistic model that simulates miRNA biogenesis to score potential miRNA candidates.
• Compatibility assessment: Assesses compatibility between the positions and frequencies of sequenced RNAs and the secondary structure of precursor molecules.
• Deep sequencing support: Leverages highly parallel sequencing technologies and deep sequencing datasets as input.
• Sensitivity and specificity: Detects small RNAs with depth and precision and distinguishes miRNAs within the bulk of sequenced transcripts.
• Validation and robustness: Demonstrated accuracy on published Caenorhabditis elegans datasets and newly generated human and dog RNA samples.
• Novel miRNA discovery and experimental validation: Reported approximately 230 novel miRNA candidates, with four new C. elegans miRNAs validated by northern blot.

Scientific Applications:

• Systematic identification: Identification of known and novel miRNAs from deep sequencing experiments.
• Functional genomics studies: Discovery of miRNAs to support studies of gene regulation, development, and disease mechanisms.
• Comparative analysis: Cross-species applications (e.g., C. elegans, human, dog) to investigate miRNA conservation and evolution.

Methodology:

Uses a probabilistic model simulating miRNA biogenesis to evaluate read position and frequency against precursor secondary structure and to score candidate miRNAs from deep sequencing data.