DHU-Pred

DHU-Pred predicts dihydrouridine (D) sites in tRNA sequences to identify post-transcriptional modification positions relevant to tRNA stability, conformational flexibility, and disease associations across bacteria, eukaryotes, and archaea.

Key Features:

• Sequence-based prediction: Predicts dihydrouridine (D) sites using tRNA sequence data.
• Feature extraction: Extracts position and composition variant features from tRNA sequences.
• Machine learning models: Applies advanced machine learning algorithms for site classification.
• Validation and performance: Assessed with evaluation metrics and experimental testing, reporting 96.9% accuracy.
• Comparative performance: Demonstrated higher accuracy than existing dihydrouridine site predictors.
• Biological relevance: Targets modifications implicated in tRNA stability, conformational flexibility, and links to pulmonary carcinogenesis and cancer.

Scientific Applications:

• Mapping tRNA modifications: Identification of potential D sites in tRNA sequences across bacteria, eukaryotes, and archaea.
• Structure–function studies: Investigation of effects of dihydrouridine on tRNA stability and conformational flexibility.
• Disease association research: Prioritization of candidate D sites for studying links to pulmonary carcinogenesis and other cancer-related processes.
• Experimental planning: Selection of high-confidence predicted sites to guide targeted mass spectrometry or mutagenesis experiments.

Methodology:

Extracts position and composition variant features from tRNA sequences and trains advanced machine learning algorithms, with performance evaluated by evaluation metrics and experimental testing yielding 96.9% accuracy.