Atmospheric radiative transfer codes are used to both predict planetary spectra and in retrieval frameworks to interpret data. We have developed two open-source retrieval frameworks, BART (Bayesian Atmospheric Radiative Transfer) and PyratBay (Python Radiative Transfer in a Bayesian Framework), to characterize exoplanetary atmospheres and assess their chemical compositions, thermal profiles and cloud structures. To adequately constrain a physically plausible atmospheric configuration, one must account for the uncertainties coming from our limited knowledge of their chemical and dynamical processes. Combining a retrieval framework (an observation-driven approach that applies a statistically robust treatment of the uncertainties) and theory-driven forward models (that provide a self-consistent insight into the physical and chemical processes at play) is a particularly promising way to accurately characterize any planetary atmosphere. We apply these models to investigate the difference between the temperature structure produced with a 3D atmospheric hydrodynamic simulation of a hot-Jupiter planet and the best-fit 1D model recovered from retrieval. In addition, we present several parametrized cloud models to describe the complex aerosol structure of the gaseous envelopes and its effect on retrieval.