GraphProt: modeling binding preferences of RNA-binding proteins.

Area of Science:

• Computational biology
• Bioinformatics
• Molecular biology

Background:

• RNA-binding proteins (RBPs) play crucial roles in gene regulation.
• Understanding RBP sequence and structure binding preferences is essential for deciphering their functions.
• High-throughput experimental data provides valuable insights into RBP-RNA interactions.

Purpose of the Study:

• To introduce GraphProt, a novel computational framework for learning RBP binding preferences.
• To validate GraphProt's accuracy and biological relevance using existing literature and experimental data.
• To demonstrate the utility of GraphProt in predicting RBP binding sites and affinities.

Main Methods:

• GraphProt utilizes high-throughput experimental data to model sequence- and structure-based binding preferences of RBPs.
• Benchmarking involves comparing GraphProt's predictions with known RBP binding characteristics from literature.
• Two applications are presented: correlating estimated binding affinities with experimental measurements and analyzing predicted Ago2 targets upon knockdown.

Main Results:

• GraphProt successfully models RBP binding preferences, aligning with established literature findings.
• Estimated binding affinities from GraphProt correlate well with experimental measurements.
• Predicted targets of Ago2 show altered expression patterns after Ago2 knockdown, validating the model's predictive power.

Conclusions:

• GraphProt is a robust computational framework for predicting RBP binding sites and affinities.
• The framework's ability to learn sequence and structure preferences enhances understanding of RBP functions.
• GraphProt models are essential tools for computational prediction in molecular biology and gene regulation studies.