CPE-SLDI

CPE-SLDI improves prediction of coding potential in RNA sequences by applying machine learning and oversampling to address local data imbalance associated with short Open Reading Frames (sORFs) (ORF length < 303 nucleotides) and distinguish coding RNAs from non-coding RNAs (ncRNAs).

Key Features:

• sORF targeting: Specifically addresses sequences containing short Open Reading Frames (sORFs) with ORF length < 303 nucleotides.
• Local data imbalance correction: Handles local imbalance between coding RNAs with sORFs and ncRNAs.
• Oversampling: Employs oversampling techniques to augment datasets for coding RNAs with sORFs.
• Machine learning prediction: Uses advanced machine learning techniques to construct a model distinguishing coding RNAs and ncRNAs.
• Sequence-derived features: Integrates various sequence-derived features from augmented datasets into the prediction model.
• Validation: Demonstrated improved coding potential prediction performance in comparative studies against existing methods.

Scientific Applications:

• Functional genomics: Improving annotation of RNA coding potential in functional genomics studies.
• sORF research: Characterizing RNAs that contain sORFs (ORF length < 303 nt) to clarify their coding potential.
• Disease mechanism exploration: Investigating RNA roles and disease associations by providing more accurate coding versus non-coding classification.

Methodology:

Uses oversampling to augment coding-RNA sORF datasets and constructs a machine learning prediction model that integrates various sequence-derived features from the augmented datasets, with performance evaluated in comparative studies.