Comparative analysis of protein-coding and long non-coding transcripts based on RNA sequence features

Area of Science:

• Molecular Biology
• Genomics
• Bioinformatics

Background:

• Long non-coding RNAs (lncRNAs) are increasingly recognized for their roles in cellular processes, yet their coding potential remains debated.
• Ribosome profiling (Ribo-seq) is a key technology for studying RNA translation, including initiation and elongation.
• Some studies suggest lncRNAs may have coding potential, challenging their 'non-coding' classification.

Purpose of the Study:

• To develop a novel computational model for predicting RNA translation start sites.
• To differentiate between human messenger RNAs (mRNAs) and lncRNAs based on translation start site characteristics.
• To investigate the presence and size of putative open reading frames (ORFs) in lncRNAs.

Main Methods:

• Utilized ribosome profiling (Ribo-seq) data to identify translation initiation events.
• Developed a position weight matrix model based on identified translation starts.
• Applied the model to discriminate between human mRNAs and lncRNAs and to predict ORFs in RNA sequences.

Main Results:

• Achieved 96% accuracy in discriminating between human mRNAs and lncRNAs using the developed model.
• The model identified that the majority of lncRNAs harbor small open reading frames (ORFs) typically less than 100 nucleotides.
• In contrast, mRNAs were found to possess larger ORFs, consistent with their known coding function.

Conclusions:

• The novel position weight matrix model effectively distinguishes protein-coding mRNAs from lncRNAs.
• Findings suggest that most lncRNAs do not encode substantial proteins due to their small ORFs.
• This research contributes to a better understanding of the functional landscape of lncRNAs and their distinction from mRNAs.