Predicting RNA structure and dynamics with deep learning and solution scattering.

Area of Science:

• Structural biology
• Computational biology
• Biophysics

Background:

• RNA molecules exhibit conformational flexibility, making it difficult to predict their structures in solution under varying conditions.
• Small-angle X-ray scattering (SAXS) is crucial for validating predicted RNA structures by comparing experimental data with calculated profiles.
• Existing methods struggle with accurately representing RNA plasticity and incorporating essential cations like Mg2+ in structural models.

Purpose of the Study:

• To develop an advanced computational tool, SCOPER (Solution Conformation Predictor for RNA), for predicting and validating RNA structures in solution.
• To address the challenges of RNA conformational plasticity and the accurate inclusion of ions in structural modeling.
• To improve the accuracy of SAXS profile fitting for RNA structures by accounting for solution conditions.

Main Methods:

• Integration of kinematics-based conformational sampling with a deep learning model (IonNet) for predicting Mg2+ ion binding sites.
• Development of a pipeline (SCOPER) to model RNA solution conformations, including ion interactions and conformational ensembles.
• Benchmarking SCOPER against 14 experimental SAXS datasets to evaluate its performance.

Main Results:

• SCOPER significantly improved the quality of SAXS profile fits by incorporating Mg2+ ions and sampling RNA conformational plasticity.
• The study demonstrated that increased ion content correlates with decreased RNA plasticity.
• Accurate adjustment of plasticity and ion density is critical to prevent overfitting experimental SAXS data.

Conclusions:

• SCOPER provides an efficient and accurate method for validating RNA solution structures and generating corrected atomistic models, including ions.
• The tool is effective when provided with an initial, sufficiently accurate RNA structure.
• Understanding the interplay between ion concentration and RNA flexibility is key for accurate structural predictions in solution.