The availability of new computers and software is making possible the theoretical representation of DNA, increasing then our knowledge on the behavior of one of the most relevant biological macromolecules. Unfortunately, current simulation procedures present two major problems, which handicapped their use: i) classical force-fields present well known biases, which limit their accuracy; ii) current atomistic procedures are limited to study systems in the range of 100 base pairs (around 34 nm long), while the DNA of the simplest prokaryotic organisms is one billion times larger. The main objective of this proposal is the development of a multiscale simulation technology for the study of DNA, which will cover, with different levels of resolution, but with the same physical roots, the entire range of DNA scales, from nucleobase (Ǻ-scale) to the human genome (m-scale). Our roadmap will start for the development of a polarized force-field which will be parametrized against a variety of experimental and theoretical data. In a second stage, we will analyze a very large number of DNA sequences in different epigenetic and packing states and we will create a MoDEL-like database of DNA trajectories. In a third stage we will derive coarse grained and essential dynamic-based strategies for ultra-fast accurate simulations for medium to long segments of DNA. In the last stage of this project we will develop a new mesoscopic model, which will go beyond the harmonic nearest-neighbors model, accounting for multi-modality, for neutralization-induced deformations, and for changes in DNA properties related to epigenetic changes. Using these models we expect to analyze fine details of (human) genome structure and regulation, trying to reach the connection point between physical properties of DNA, chromatine structure, epigenetic signatures and gene regulation