A practical approach for addressing the computer simulation of protein-carbohydrate interactions is described here. An articulated computational protocol was setup and validated by checking its ability to predict experimental data, available in theliterature, and concerning the selectivity shown by the Carbohydrate Recognition Domain(CRD) of the human asialoglycoprotein receptor (ASGP-R) toward Gal-type ligands. Somerequired features responsible for the interactions were identified. Subsequently the sameprotocol was applied to monomer sugar molecules that constitute the building blocks foralginates and ulvans. Such sugar polymers may supply a low-cost source of rare sugars witha potential impact on several industrial applications, from pharmaceutical to fine chemicalindustry. An example of their applicative exploitation could be given by their use indeveloping biomaterial with adhesion properties toward hepatocytes, through interactionwith the ASGP-R. Such a receptor has been already proposed as a target for exogenousmolecules, specifically in the case of hepatocytes, for diagnostic and therapeutic purposes.The DOCK5.2 program was used to search optimal locations of the above ligands of interestinto CRD binding site and to roughly estimate interaction energies. Finally, the binding ∆G oftheoretical protein-ligand complexes was estimated by using the DelPhi program in which thesolvation free energy is accounted for with a continuum solvent model, by solving the Poisson-Boltzmann equation. The structure analysis of the obtained complexes and their ∆G values suggest that one of the sugar monomers of interest shows the desired characteristics. Full article

Topological indices have been applied to build QSAR models for a set of 20 an-timalarial cyclic peroxy cetals. In order to evaluate the reliability of the proposed linearmodels leave-n-out and Internal Test Sets (ITS) approaches have been considered. The pro-posed procedure resulted in a robust and consensued prediction equation and here it isshown why it is superior to the employed standard cross-validation algorithms involvingmultilinear regression models. Full article

This review summarises the state-of-the-art methodologies used for designing homogeneous catalysts and optimising reaction conditions (e.g. choosing the right solvent). We focus on computational techniques that can complement the current advances in high-throughput experimentation, covering the literature in the period 1996-2006. The review assesses the use of molecular modelling tools, from descriptor models based on semiempirical and molecular mechanics calculations, to 2D topological descriptors and graph theory methods. Different techniques are compared based on their computational and time cost, output level, problem relevance and viability. We also review the application of various data mining tools, including artificial neural networks, linear regression, and classification trees. The future of homogeneous catalysis discovery and optimisation is discussed in the light of these developments. Full article