ScanFold 2.0: a rapid approach for identifying potential structured RNA targets in genomes and transcriptomes

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Identifying functional RNA elements is crucial for basic research and drug discovery.
• The ScanFold algorithm maps RNA structural stability and identifies sequence-ordered structures.
• Predicting secondary structural stability for randomized sequences is computationally intensive.

Purpose of the Study:

• To overcome the speed limitations of the original ScanFold algorithm for analyzing large genomic and transcriptomic datasets.
• To introduce ScanFold 2.0, a revised algorithm that significantly enhances the speed of RNA structure prediction.

Main Methods:

• ScanFold 2.0 estimates randomized sequence folding energy using a machine learning approach, replacing explicit evaluation.
• The revised algorithm allows for the analysis of larger sequences and integration of computationally expensive folding algorithms.
• Performance was evaluated by re-analyzing Zika, HIV, and SARS-CoV-2 genomes and comparing results with ScanFold 1.0.

Main Results:

• ScanFold 2.0 achieves prediction speed increases of over 100-fold for high randomization numbers compared to ScanFold 1.0.
• The study validated ScanFold 2.0's predictions against biochemical structure probing datasets for SARS-CoV-2.
• Consistent results were observed between ScanFold 1.0 and ScanFold 2.0, with significant time savings for the latter.

Conclusions:

• ScanFold 2.0 offers a substantial speed improvement for identifying RNA structural elements in large-scale datasets.
• The machine learning approach in ScanFold 2.0 makes RNA structure prediction more efficient and accessible.
• This advancement facilitates broader application in target discovery and therapeutic intervention research.