CNIT: a fast and accurate web tool for identifying protein-coding and long non-coding transcripts based on intrinsic...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• High-throughput sequencing generates vast amounts of data.
• Classifying RNA transcripts as protein-coding or non-coding remains challenging, particularly for poorly annotated species.

Purpose of the Study:

• To introduce an upgraded tool, the Coding-Non-Coding Identifying Tool (CNIT).
• To provide a faster and more accurate method for evaluating the coding potential of RNA transcripts.

Main Methods:

• Development of CNIT, an upgraded version of the original Coding-Non-Coding Index (CNCI).
• Performance evaluation of CNIT against CNCI using various species, including human, mouse, zebrafish, fruit fly, worm, and Arabidopsis.
• Assessment of CNIT's accuracy using Area Under the Curve (AUC) values across 11 animal and 27 plant species.

Main Results:

• CNIT demonstrates significantly improved performance, running approximately 200 times faster than CNCI.
• CNIT exhibits higher accuracy in classifying RNA transcripts across multiple species compared to CNCI.
• CNIT achieves high accuracy for nearly all eukaryotic transcripts, as indicated by AUC values.

Conclusions:

• CNIT offers a substantial advancement in the computational identification of coding and non-coding RNA transcripts.
• The tool's speed and accuracy make it valuable for analyzing large-scale transcriptomic data.
• A publicly accessible, mobile-friendly web server for CNIT is available for research use.