ncRNA Coding Potential Prediction Using BiLSTM and Transformer Encoder-Based Model.

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Noncoding RNAs (ncRNAs) play crucial roles in gene regulation.
• Emerging evidence shows ncRNAs can be translated into functional peptides.
• Existing computational tools primarily distinguish ncRNAs from messenger RNAs (mRNAs).

Purpose of the Study:

• To develop a highly accurate computational tool for predicting the coding potential of ncRNAs.
• To facilitate the discovery of novel peptides derived from ncRNAs.
• To improve the investigation of ncRNA functions.

Main Methods:

• Proposed a novel BiLSTM and Transformer encoder-based model (nBAT).
• Incorporated a learnable position encoding mechanism for ncRNA sequence embeddings.
• Extracted and encoded 43 intrinsic features into the Transformer encoder.

Main Results:

• nBAT demonstrated superior performance compared to state-of-the-art methods on various datasets.
• The model achieved competitive performance when applied to new ncRNAs.
• nBAT effectively predicts the coding potential of ncRNAs.

Conclusions:

• nBAT is a highly accurate tool for predicting ncRNA coding potential.
• The model aids in the discovery of novel peptides from ncRNAs.
• nBAT can be utilized for high-throughput coding potential prediction of ncRNAs.