epl draft (accepted for publication) Finite enthalpy model parameters from DNA melting tempera- tures This is an author’s post-print of EPL (Europhysics Letters) 12/2011; 96(6). DOI:10.1209/0295-5075/96/68001 Gerald Weber Department of Physics, Federal University of Minas Gerais, 31270-901 Belo Horizonte-MG, Brazil PACS 87.14.gk – Biomolecules: types: DNA PACS 87.15.Zg – Biomolecules: structure and physical properties: Phase transitions PACS 87.15.A- – Biomolecules: structure and physical properties: Theory, modeling, and com- puter simulation Abstract – Peyrard-Bishop (PB) models are used for the study of denaturation in DNA. Un- fortunately, there is little connection of these models to linear nearest neighbour models which are extensively used for the calculation of melting temperatures in biochemistry. Here we use the Joyeux-Buyukdagli (JB) model, a variant of the PB model which incorporates stacking enthalpies, and carry out a fitting procedure to experimental melting temperatures where we let the enthalpies vary freely. We start out with a single value for enthalpy for all combination of base pairs and after the fitting we obtain a new set of enthalpies which correlate very strongly with measured enthalpies. This result provides the needed support for the use of experimental enthalpies in the JB/PB model. One of the earliest approaches to predict experimen- tal melting temperatures are first neighbour or nearest neighbour (NN) models [1–4]. In this approach, the dou- ble stranded DNA sequence is broken up into units of se- quential neighbouring base pairs. To each of these nearest neighbours one assigns a value for entropy and enthalpy representing the stability of the molecule. This results into a simple linear model and the entropy and enthalpy are easily obtained from melting temperatures with the use of standard numerical techniques. The resulting param- eters can then be used to predict melting temperatures for unknown DNA sequences and its good accuracy has made it a very popular method. The NN models, how- ever, do not provide much insight into what is happening to the DNA molecule. For instance, they are of limited use when it comes to understand local base-pair opening during denaturation. Fortunately, the denaturation of DNA can be modelled by a large number of theoretical techniques, ranging from Ising type Hamiltonians [5] to atomistic molecular dynam- ics [6]. One important model, the Peyrard-Bishop model (PB) [7], finds a compromise between computational ef- ficiency and physics by employing a Hamiltonian which accounts for the basic ingredients of the stability of the DNA molecule: the hydrogen bonds and the stacking in- teraction. It can be used to calculate the average open- ings of the double strand from an equilibrium partition function which can be correlated to melting temperatures of DNA [8]. PB models have found their use in numer- ous applications, for instance they were recently applied to the study of thermal transport in DNA [9], genomic melting [10], pre-melting dynamics of DNA [11], to the analysis of localisation in DNA [12], and to study the or- der of the denaturation transition [13]. Please note that this list highlights just a few recent applications and is by no means exhaustive. Given the broad application of the model, there is an active interest in pursuing modi- fications to the Hamiltonian to cover a range of impor- tant properties of the DNA molecule. For instance, it is possible to add solvation barriers [14, 15] and other modi- fied potentials [16]. Using this model the thermodynamic properties of oligonucleotides can be studied with a vari- ety of theoretical techniques such as wavelet analysis [17], Langevin [18] or Fokker-Plank formalism [19], again just to cite but a few examples. Unfortunately, the PB model suffers from a chronic lack of realistic parameters. One of the earliest works towards realistic parameters, and still one of the most frequently p-1