Developing an Updated Strategy for Estimating the Free-Energy Parameters in RNA Duplexes

Area of Science:

• Biophysics
• Computational Biology
• RNA Structure

Background:

• Current RNA duplex free energy models rely on dinucleotide parameters and constants, with benchmark data biased towards shorter sequences (6-8 base pairs).
• Existing models may not accurately represent longer RNA duplexes due to limitations in benchmark data length and composition.

Purpose of the Study:

• To investigate if alternative models, including longer RNA duplexes (17 base pairs), yield different thermodynamic parameters.
• To develop and test an updated strategy for fitting experimental melting curve data to obtain RNA duplex parameters.

Main Methods:

• Developed a gradient-descent fitting program to directly extract thermodynamic parameters (ΔG, ΔH, ΔS, Tm) from experimental melting curves.
• Combined gradient descent with a genetic algorithm to integrate new duplex melting results with standard benchmark data.
• Tested both the standard Turner model and a novel model incorporating length-dependent terms.

Main Results:

• Both the standard and new models could fit the existing benchmark data.
• The new model demonstrated superior performance in handling and accurately parameterizing longer RNA sequences.
• An updated strategy for fitting duplex melting data was successfully developed.

Conclusions:

• The inclusion of longer RNA duplexes and length-dependent terms in thermodynamic models improves accuracy.
• The developed gradient-descent fitting strategy offers a more robust method for RNA parameterization.
• This work refines the understanding and prediction of RNA duplex stability.