Computational identification of protein binding sites on RNAs using high-throughput RNA structure-probing data

Area of Science:

• Computational Biology
• Molecular Biology
• Genomics

Background:

• High-throughput sequencing of structure-probing data offers insights into RNA properties.
• Analyzing these complex datasets requires sophisticated statistical modeling.

Purpose of the Study:

• To develop and validate statistical methods for extracting structural features from RNA structure-probing data.
• To demonstrate the utility of these features in predicting RNA functional elements and protein binding sites.

Main Methods:

• Statistical modeling of RNA structure-probing data.
• Feature extraction from probing data.
• Machine learning for prediction of RNA zipcodes and protein binding sites.

Main Results:

• Developed methods for statistically modeling RNA structure-probing data.
• Extracted features accurately predict RNA zipcodes in yeast.
• Demonstrated improved prediction of RNA binding protein sites from gPAR-CLIP data compared to raw data or sequence features.

Conclusions:

• Statistical modeling of RNA structure-probing data provides valuable insights into RNA function.
• The developed methods enhance the prediction of key RNA structural and functional elements.
• This approach advances the understanding of RNA regulation and function.