We introduce PHASE, a highly flexible system for common pharmacophore identification and assessment, 3D QSAR model development, and 3D database creation and searching. The primary workflows and tasks supported by PHASE are described, and details of the underlying scientific methodologies are provided. Using results from previously published investigations, PHASE is compared directly to other ligand-based software for its ability to identify target pharmacophores, rationalize structure-activity data, and predict activities of external compounds.

Pharmacophore modeling and 3D database searching are now recognized as integral components of lead discovery and lead optimization, and the continuing need for improved pharmacophorebased tools has driven the development of PHASE. By employing a novel, tree-based partitioning algorithm, PHASE exhaustively identifies spatial arrangements of functional groups that are common and essential to the biologic activity of a set of high affinity ligands. These pharmacophore hypotheses are validated in a number of ways, including their ability to: (i) rationalize the binding affinities of a training set of molecules of varying activity, (ii) successfully predict the affinities of a test set of molecules, and (iii) selectively retrieve known actives from a database of drug-like molecules. In addition, PHASE uniquely offers the ability to distinguish multiple binding modes through a bi-directional clustering approach applied to bit string representations of the ligand/hypothesis space.Pharmacophore Perception and 3D Quantitative Structure-activity relationship (QSAR) Development A collection of 49 glycoprotein (GP)IIb/IIIa antagonists (RGD mimics) of varying affinity and spanning two distinct chemotypes was used to define a 23-member training set and a 26-member test set for the generation and validation of 3D pharmacophore models that rationalize the associated fibrinogen receptor-binding data (1,2). Activity thresholds of 100 nM and 1 lM were used to identify nine actives and five inactives, respectively, within the training set. A maximum of 500 conformations were generated for each molecule using MacroModel torsional sampling with OPLS_2005 postprocessing (MacroModel 9.1 Reference Manual, Copyright ª 2005, Schrçdinger, L.L.C., New York, USA; http://www.schrodinger.com/). Pharmacophores with four features (including positive and negative ionic centers) common to all nine training set actives were identified then scored according to superposition of pharmacophore site points, alignment of vector characteristics, overlap of molecular

By using molecular dynamics simulation technique we studied the changes occurring in membranes constructed of dipalmitoylphosphatidylcholine (DPPC) and cholesterol at 8:1 and 1:1 ratios. We tested two different initial arrangements of cholesterol molecules for a 1:1 ratio. The main difference between two initial structures is the average number of nearest-neighbor DPPC molecules around the cholesterol molecule. Our simulations were performed at constant temperature (T = 50 degrees C) and pressure (P = 0 atm). Durations of the runs were 2 ns. The structure of the DPPC/cholesterol membrane was characterized by calculating the order parameter profiles for the hydrocarbon chains, atom distributions, average number of gauche defects, and membrane dipole potentials. We found that adding cholesterol to membranes results in a condensing effect: the average area of membrane becomes smaller, hydrocarbon chains of DPPC have higher order, and the probability of gauche defects in DPPC tails is lower. Our results are in agreement with the data available from experiments.

Five molecular dynamics computer simulations were performed on different phospholipid:sterol membrane systems in order to study the influence of sterol structure on membrane properties. Three of these simulated bilayer systems were composed of a 1:8 sterol:phospholipid ratio, each of which employed one of the sterol molecules: cholesterol, ergosterol, and lanosterol. The two other simulations were of a bilayer with a 1:1 sterol:phospholipid ratio. These simulations employed cholesterol and lanosterol, respectively, as their sterol components. The observed differences in simulations with cholesterol and lanosterol may have their implication on the form of the phospholipid/sterol phase diagram.

The structural and dynamical properties of a solvated proton in the influenza A virus M2 channel are studied using a molecular dynamics (MD) simulation technique. The second-generation multi-state empirical valence bond (MS-EVB2) model was used to describe the interaction between the excess proton and the channel environment. Solvation structures of the excess proton and its mobility characteristics along the channel were determined. It was found that the excess proton is capable of crossing the channel gate formed by the ring of four histidine residues even though the gate was only partially open. Although the hydronium ion itself did not cross the channel gate by traditional diffusion, the excess proton was able to transport through the ring of histidine residues by hopping between two water molecules located at the opposite sides of the gate. Our data also indicate that the proton diffusion through the channel may be correlated with the changes in channel conformations. To validate this observation, a separate simulation of the proton in a "frozen" channel has been conducted, which showed that the proton mobility becomes inhibited.