Parameterization and Conformational Sampling Effects in Pharmacophore Multiplet Searching

A new 4-point pharmacophore method for molecular similarity and diversity that rapidly calculates all potential pharmacophores/pharmacophoric shapes for a molecule or a protein site is described. The method, an extension to the ChemDiverse/Chem-X software (Oxford Molecular, Oxford, England), has also been customized to enable a new internally referenced measure of pharmacophore diversity. The "privileged" substructure concept for the design of high-affinity ligands is presented, and an example of this new method is described for the design of combinatorial libraries for 7-transmembrane G-protein-coupled receptor targets, where "privileged" substructures are used as special features to internally reference the pharmacophoric shapes. Up to 7 features and 15 distance ranges are considered, giving up to 350 million potential 4-point 3D pharmacophores/molecule. The resultant pharmacophore "key" ("fingerprint") serves as a powerful measure for diversity or similarity, calculable for both a ligand and a protein site, and provides a consistent frame of reference for comparing molecules, sets of molecules, and protein sites. Explicit "on-the-fly" conformational sampling is performed for a molecule to enable the calculation of all geometries accessible for all combinations of four features (i.e., 4-point pharmacophores) at any desired sampling resolution. For a protein site, complementary site points to groups displayed in the site are generated and all combinations of four site points are considered. In this paper we report (i) the details of our customized implementation of the method and its modification to systematically measure 4-point pharmacophores relative to a "special" substructure of interest present in the molecules under study; (ii) comparisons of 3- and 4-point pharmacophore methods, highlighting the much increased resolution of the 4-point method; (iii) applications of the 4-point potential pharmacophore descriptors as a new measure of molecular similarity and diversity and for the design of focused/biased combinatorial libraries.

Electron-electron repulsion energy (<Vee>) is presented as a new molecular descriptor to be employed in QSAR and QSPR studies. Here it is shown that this electronic energy parameter is connected to molecular quantum similarity measures (MQSM), and as a consequence can be considered as a complement to steric and electronic parameters in description of molecular properties and biological responses of organic compounds. The present strategy considers the molecule as a whole, thus there is no need to employ contributions of isolated fragments as in many calculations of molecular descriptors, like log P or the Free-Wilson analysis. The procedure has been tested in a widespread set of molecules: alcohols, alkanamides, indole derivatives and 1-alkylimidazoles. Molecular properties, as well as toxicity, are correlated using <Vee> as a parameter, and extensions to the method are given for handling difficult systems. In almost all studied cases, satisfactory linear relationships were finally obtained.

The combination of chromatography and chemometrics has been introduced in order to provide information about drug analytes, biological macromolecules, the stationary phase, whose properties are related to molecular pharmacology and rational drug design. In this work a pharmacological classification was made in order to find strategy for predicting an activity of anticancer drugs. Principal Component Analysis (PCA) method has been employed to build some relationship models between the following: lipophilicity parameters of selected antitumor drugs obtained by chromatographic analysis, some molecular modeling descriptors from HyperChem software, lipophilicity parameters calculated by proper program, biological activity, as well as mechanism of actions and therapeutic properties of selected anticancer drugs. The most significant influence on factors value had parameters, which described compounds’ lipophilicity. Important debt on antitumor properties had some other molecular descriptors, which were considered in this paper (e.g. topological, physico-chemical, and energy ones). In addition, distribution of individual drugs on plots determined by two principal components had shown points in good accordance with their chemical structures and cytostatic activity as well as their mechanism of actions and therapeutic properties.

The increasing main-memory capacity has allowed query execution to occur primarily in main memory. Database systems employ compression, not only to fit the data in main memory, but also to address the memory bandwidth bottleneck. Lightweight compression schemes focus on efficiency over compression rate and allow query operators to process the data in compressed form. For instance, dictionary compression keeps the distinct column values in a sorted dictionary and stores the values as index codes with the minimum number of bits. Packing the bits of each code contiguously, namely horizontal bit packing, has been optimized by using SIMD instructions for unpacking and by evaluating predicates in parallel per processor word for selection scans. Interleaving the bits of codes, namely vertical bit packing, provides faster scans, but incurs prohibitive costs for packing and unpacking. Here, we improve packing and unpacking for vertical bit packing using SIMD instructions, achieving more than an order of magnitude speedup. Also, we optimize horizontal bit packing on the latest CPUs and compare all approaches. While no single variant is better in all cases, vertical bit packing offers a good trade-off by combining the fastest scans with comparably fast packing and unpacking.

Communication bottleneck has been identified as a significant issue in distributed optimization of large-scale learning models. Recently, several approaches to mitigate this problem have been proposed, including different forms of gradient compression or computing local models and mixing them iteratively. In this paper, we propose Qsparse-local-SGD algorithm, which combines aggressive sparsification with quantization and local computation along with error compensation, by keeping track of the difference between the true and compressed gradients. We propose both synchronous and asynchronous implementations of Qsparse-local-SGD . We analyze convergence for Qsparse-local-SGD in the distributed setting for smooth non-convex and convex objective functions. We demonstrate that Qsparse-local-SGD converges at the same rate as vanilla distributed SGD for many important classes of sparsifiers and quantizers. We use Qsparse-local-SGD to train ResNet-50 on ImageNet and show that it results in significant savings over the state-of-the-art, in the number of bits transmitted to reach target accuracy.

This chapter deals with the tendencies in design of multivalent neoglycoconjugates for glycobiology research and high-throughput profiling technologies, including cellular phenotyping. Soluble polyacrylamide (PAA) conjugates are remarkable owing to a variety of possibilities for synthesis and application. PAA is soluble and stable, and the molecule is flexible, PAA-tethered ligands are capable of adjusting to a receptor during multiple-point interactions and PAA does not bind to the cell surface. Synthesis provides unlimited diversity of the probe types (biotin, fluorescein, allyl, digoxygenin, 3H, radiolabelled I), glyco-particles, glyco-surfaces, multiarrays, immunogens etc. Several examples illustrate the most advanced applications. (i) Dynamic systems: the selectin ligands immobilized on the surface as sugar-PAA conjugates made the study of the kinetics of rolling in the model system possible. Carbohydrate ligands that are covalently attached to the chip as sugar-PAA conjugates are of use with the surface plasmon resonance method. (ii) Pseudoglycoprotein: some questions arise regarding the biologically active glycoproteins, e.g.: is a carbohydrate or peptide fragment responsible for the activity? We have proposed the approach that promotes to answer this and other questions. The pool of oligosaccharides that were spitted off a glycoprotein is attached to PAA resulting in a pseudoglycoprotein. (iii) Virtual (dynamic) glycotope: receptor-ligand recognition, such as that of P-selectin with its receptor P-selectin glycoprotein ligand 1, frequently involves molecular interactions at two distinct sites. Using P-selectin as a model, we developed an approach to discover novel ligands. PAA was synthesized with multiple ligands; a marked synergistic inhibitory effect was observed.

In principle, for the development of pharmacophore models, two different approaches are known, the ligand-based and the structure-based approaches. Ligand-based pharmacophore modeling is an important tool for facilitating drug discovery in the absence of a macromolecular target structure. Structure-based pharmacophore modeling is based on the 3D structure of a macromolecule-ligand complex or a macromolecule without a ligand. As every virtual screening tool, also pharmacophore modeling very much depends on the data employed for model generation, refinement, and theoretical validation. This chapter describes the entire multiple ligand continuum from linked pharmacophore patterns to highly integrated examples including biological results to illustrate the broad spectrum of successful pathways for design and development of multiple ligands. It discusses the author's work to develop efficient kinase inhibitors to enable the development and characterization of further kinase targets and drugs.

Natural product and synthetic macrocycles are chemically and topologically diverse. An efficient, accurate, and general method for generating macrocycle conformations would enable structure-based design of macrocycle drugs or host-guest complexes. Computational sampling also provides insight into transiently populated states, complementing crystallographic and NMR data. Here, we report a new algorithm, BRIKARD, that addresses this challenge through computational algebraic geometry and inverse kinematics together with local energy minimization. BRIKARD is demonstrated on 67 diverse macrocycles with structural data, encompassing various ring topologies. We find this approach enumerates diverse structures with macrocyclic RMSD < 1.0 Å to the experimental conformation for 85% of our data set in contrast to success rates of 67-75% with other approaches, while for the subset of 21 more challenging compounds in the data set, these rates are 57% and 10-29%, respectively. Because the algorithm can be efficiently run in parallel on many processors, exhaustive conformational sampling of complex cycles can be obtained in minutes rather than hours: with a 40 processor implementation protocol, BRIKARD samples the conformational diversity of a potential energy landscape in a median of 1.3 minutes of wallclock time, much faster than 3.1-10.3 hours necessary with current programs. By rigorously testing BRIKARD on a broad range of scaffolds with highly complex ring systems, we push the frontiers of macrocycle sampling to encompass multiring compounds, including those with more than 50 ring atoms and up to seven interlaced flexible rings.

We consider a generalized-ensemble algorithm for coarse-grained simulations of biomolecules which allows the thermodynamic behavior of two or more sequences to be determined in a single multisequence run. By carrying out a random walk in sequence space, the method also enhances conformational sampling. Escape from local energy minima is accelerated by visiting sequences for which the minima are more shallow or absent. We test the method on an intermediate-resolution coarse-grained model for protein folding with 3 amino acid types and explore the potential for a large-scale coverage of sequence space by applying the method to sets of more than 1000 sequences. The resulting thermodynamic data are used to analyze the structures and stability properties of sequences covering the space between folds with different secondary structures.

We will summarize the latest developments for our protein-ligand docking software PLANTS (Protein-Ligand ANT System) [1][2][3]. This will include the pre-processing of the ligands and the protein performed by the program SPORES (Structure PrOtonation and REcognition System), parameterization of a new scoring function, as well as the inclusion of additional degrees of freedom into the docking process like flexible side chains and essential water molecules. The use of constraints from experimental data for improving the docking poses will be presented. Finally, an outlook towards the possible usage of the PLANTS approach for the flexible alignment of multiple ligands will be given.

3-D ligand conformations are required for most ligand-based drug design methods, such as pharmacophore modeling, shape-based screening, and 3-D QSAR model building. Many studies of conformational search methods have focused on the reproduction of crystal structures (i.e. bioactive conformations); however, for ligand-based modeling the key question is how to generate a ligand alignment that produces the best results for a given query molecule. In this work, we study different conformation generation modes of ConfGen and the impact on virtual screening (Shape Screening and e-Pharmacophore) and QSAR predictions (atom-based and field-based). In addition, we develop a new search method, called common scaffold alignment, that automatically detects the maximum common scaffold between each screening molecule and the query to ensure identical coordinates of the common core, thereby minimizing the noise introduced by analogous parts of the molecules. In general, we find that virtual screening results are relatively insensitive to the conformational search protocol; hence, a conformational search method that generates fewer conformations could be considered "better" because it is more computationally efficient for screening. However, for 3-D QSAR modeling we find that more thorough conformational sampling tends to produce better QSAR predictions. In addition, significant improvements in QSAR predictions are obtained with the common scaffold alignment protocol developed in this work, which focuses conformational sampling on parts of the molecules that are not part of the common scaffold.

What makes an agonist and a competitive antagonist? Inthis work,we aim to answer this question by performing parallel tempering MonteCarlo simulations on the serotonin type 3A (5-HT3A) receptor.We use linear response theory to predict conformational changes inthe 5-HT3A receptor active site after weak perturbationsare applied to its allosteric binding sites. A covariance tensor isbuilt from conformational sampling of its apo state, and a harmonicapproximation allows us to substitute the calculation of ligand-inducedforces with the binding site’s displacement vector. Remarkably,our study demonstrates the feasibility of effectively discerning betweenagonists and competitive antagonists for multiple ligands, requiringcomputationally expensive calculations only once per protein.