Replacing salt correction factors with optimized RNA nearest-neighbour enthalpy and entropy parameters

Abstract

We calculate the nearest-neighbour enthalpies and entropies at 5 salt concentrations of 18 RNA sequences, each for at least 9 different species concentrations, totalling 757 melting temperatures, using a melting temperature optimization method. These new parameters do not need to be saltcorrected and are shown to provide overall improved melting temperature predictions. They show a marked quadratic dependence with salt concentrations which are compensated to form linear Gibbs free energies. Two different parameter schemes were tested, with fixed or variable initial parameters. We have found that using variable initial parameters provides better predictive results than using salt correction factors and that the prediction uncertainty is considerably reduced for a validation set of independent sequences. An interpolation scheme is introduced to generate model parameters for arbitrary salt concentrations which performs better against a validation set than predictions using salt corrections.

Salt Dependent Mesoscopic Model for RNA with Multiple Strand Concentrations

“…41,49 Nearest-neighbour (NN) models are typically limited to temperature prediction, and terminal effects are included as an energy penalty. 50 The salt dependence of these terminal factors were studied by us recently, 51 and we observed a marked quadratic dependence in the enthalpies and entropies with salt concentration which are compensated to form almost linear Gibbs free energies. 51 Here, we adapt the mesoscopic PB model to RNA with varying salt dependence, multiple strand concentrations and including terminal effects.…”

“…50 The salt dependence of these terminal factors were studied by us recently, 51 and we observed a marked quadratic dependence in the enthalpies and entropies with salt concentration which are compensated to form almost linear Gibbs free energies. 51 Here, we adapt the mesoscopic PB model to RNA with varying salt dependence, multiple strand concentrations and including terminal effects. In part we applied a similar approach as from our previous work on DNA salt-dependent terminal effects, 52 which enables us to compare RNA and DNA terminal effects and discuss their differences.…”

“…We used the set of RNA melting temperatures from Ferreira et al 51 , consisting of 18 RNA duplexes at four different [Na + ] concentrations (71, 121, 221, 621 mM). For each sequence and salt concentrations there are at least 9 measurements at different strand concentrations C t in the range of 5 µM to 700 µM, see Fig.…”

“…1. This does not represent a problem for the nearest-neighbour model, 51 but for the PB model it becomes necessary to group the melting temperatures together to the closest value of C t .…”

“…1. 51 This represents a challenge for the mesoscopic model which usually requires that all temperatures are at the same concentration. 53 The reason for this is that the PB model is a single molecule calculation, and the melting temperatures are correlated to experimental values at a single strand concentration.…”