Upperbound procedures for the identification of similar three-dimensional chemical structures

Summary This paper describes techniques for calculating the degree of similarity between an input query molecule and each of the molecules in a database of 3-D chemical structures. The inter-molecular similarity measure used is the number of atoms in the 3-D common substructure (CS) between the two molecules which are being compared. The identification of 3-D CSs is very demanding of computational resources, even when an efficient clique detection algorithm is used for this purpose. Two types of upperbound calculation are described which allow reductions in the number of exact CS searches which need to be carried out to identify those molecules from a database which are similar to a 3-D target molecule.     

An inequality for 3D database searching and its use in evaluating the treatment of conformational flexibility

Summary A mathematical formula is introduced for predicting the number of hits that should be observed in a flexible 3D database search, based on the results of a set of related queries. The projected number of hits is always greater than or equal to the actual number of hits, the discrepancy being due to imperfect treatment of conformational flexibility of the molecules. Hence, the difference between the projected and actual number of hits, , serves to measure how well conformational flexibility is being treated, in a manner that is objective, easy for a user to quickly verify, and independent of the particular algorithm for flexible 3D database search. It is shown that is a function both of how well conformational flexibility is treated and of the precision of the query. When the distance constraint is defined only to a precision of ±2.0 Å, in a single-conformer database of drug-like molecules values of only 0.03 are found, while in a single-conformer database of di- and tripeptides, is 0.15. At increased precision, a flexible 3D database search becomes critical. For a single-conformer database, using a query of precision ±0.2 Å, applied to a database of drug-like molecules, is 0.97; applied to a database of di- and tripeptides, is 2.21. By contrast, treating conformational flexibility by storing up to 100 conformers per molecule, at this precision, applied to a database of drug-like molecules, is 0.002; applied to a database of di- and tripeptides, is 0.07. This inequality, and hence , is defined only for database queries containing a single distance constraint; how the inequality may generalize to higher-dimensional queries is still unclear.     

A hybrid approach for addressing ring flexibility in 3D database searching

A hybrid approach for flexible 3D database searching is presented that addresses the problemof ring flexibility. It combines the explicit storage of up to 25 multiple conformations ofrings, with up to eight atoms, generated by the 3D structure generator CORINA with thepower of a torsional fitting technique implemented in the 3D database system UNITY. Acomparison with the original UNITY approach, using a database with about 130,000 entriesand five different pharmacophore queries, was performed. The hybrid approach scored, on anaverage, 10–20% more hits than the reference run. Moreover, specific problems withunrealistic hit geometries produced by the original approach can be excluded. In addition, theinfluence of the maximum number of ring conformations per molecule was investigated. Anoptimal number of 10 conformations per molecule is recommended.     

ALADDIN: An integrated tool for computer-assisted molecular design and pharmacophore recognition from geometric,steric, and substructure searching of three-dimensional molecular structures

Summary ALADDIN is a computer program for the design or recognition of compounds that meet geometric, steric, and substructural criteria. ALADDIN searches a database of three-dimensional structures, marks atoms that meet substructural criteria, evaluates geometric criteria, and prepares a number of files that are input for molecular modification and coordinate generation as well as for molecular graphics. Properties calculated from the three-dimensional structure are described by either properties calculated from the molecule itself or from the molecule as compared to a reference molecule and associated surfaces. ALADDIN was used to design analogues to probe a bioactive conformation of a small molecule and a peptide, to test alternative superposition rules for receptor mapping of the D2 dopamine receptor, to recognize unexpected D2 dopamine agonist activity of existing compounds, and to design compounds to fit a binding site on a protein of known structure. We have found that series designed by ALADDIN show much more subtle variation in shape than do those designed by traditional methods and that compounds can be designed to be very close matches to the objective.     

Fast similarity search in three-dimensional structure databases

Given a database D of three-dimensional (3D) molecular structures and a target molecule Q, the similarity search problem is to find the molecules O in D that match Q after allowing for an arbitrary number of whole-structure rotations and translations as well as a certain number of edit operations. The edit operations include relabeling an atom, deleting an atom, and inserting an atom. This search operation arises in many biochemical applications. In this paper we study the similarity search problem and a class of related queries. We present a computer vision based technique, called geometric hashing, for processing these queries. Experimental results on a database of 3D molecular structures obtained from the National Cancer Institute indicate the good performance of the presented technique.     

CLIP: similarity searching of 3D databases using clique detection

This paper describes a program for 3D similarity searching, called CLIP (for Candidate Ligand Identification Program), that uses the Bron-Kerbosch clique detection algorithm to find those structures in a file that have large structures in common with a target structure. Structures are characterized by the geometric arrangement of pharmacophore points and the similarity between two structures calculated using modifications of the Simpson and Tanimoto association coefficients. This modification takes into account the fact that a distance tolerance is required to ensure that pairs of interatomic distances can be regarded as equivalent during the clique-construction stage of the matching algorithm. Experiments with HIV assay data demonstrate the effectiveness and the efficiency of this approach to virtual screening.     

Development of an internal searching algorithm for parameterization of the MM2/MM3 force fields

Application of Allinger's MM2/MM3 force fields to molecules of real interest is frequently hindered by the lack of parameters for various heterocyclic systems and for poly-functionalized molecules. A common approach to this problem is to manually choose missing parameters “by analogy” with those that are part of the force field's internal parameter set. Naturally, this is generally attempted only by those possessing extensive experience with force fields. In order to use the MM2/MM3 force fields to study herbicides, an algorithm has been developed to automate this process for the non MM2 specialist. Using a set of “relative cost” criteria for atom type replacement, the algorithm searches the force field parameter set and selects the most appropriate parameters for a given molecule whose MM2 output file contains “missing parameter” errors. The program selects parameter error messages from a standard MM2 output file, finds analogous parameters, asks the user to verify their appropriateness and creates a standard MM2 parameter deck for the molecule of interest.     

DOCK 4.0: Search strategies for automated molecular docking of flexible molecule databases

In this paper we describe the search strategies developed for docking flexible molecules to macomolecular sites that are incorporated into the widely distributed DOCK software, version 4.0. The search strategies include incremental construction and random conformation search and utilize the existing Coulombic and Lennard-Jones grid-based scoring function. The incremental construction strategy is tested with a panel of 15 crystallographic testcases, created from 12 unique complexes whose ligands vary in size and flexibility. For all testcases, at least one docked position is generated within 2 Å of the crystallographic position. For 7 of 15 testcases, the top scoring position is also within 2 Å of the crystallographic position. The algorithm is fast enough to successfully dock a few testcases within seconds and most within 100 s. The incremental construction and the random search strategy are evaluated as database docking techniques with a database of 51 molecules docked to two of the crystallographic testcases. Incremental construction outperforms random search and is fast enough to reliably rank the database of compounds within 15 s per molecule on an SGI R10000 cpu.     

Improved intermolecular force field for crystalline oxohydrocarbons including O?HO hydrogen bonding

An improved intermolecular force field of the (exp‐6‐1) type was obtained by fitting a training set of 124 observed oxohydrocarbon crystal structures and seven observed heats of sublimation. All of these structures, when energy minimized, showed cell edge length shifts of 3% or less. Especially good results were obtained for modeling carbohydrate crystal structures. The fitted crystal structures are a subset of a database of 180 structures systematically selected from the CSD library according to functional group and threshold accuracy criteria. Energy minimization results are also presented for 56 structures not in the training set, which showed cell edge length shifts larger than 3%. The previously published W99 force field, using C(4), C(3), and H(1) potentials, was slightly modified and adopted for hydrocarbon portions of the molecules. Oxygen atoms with one bond, O(1), and those with two bonds, O(2), were assigned separate parameters. Hydrogen atoms bonded to oxygen were assigned exponential repulsion functions and divided into two types: in hydroxy groups, H(2); and in carboxyl groups, H(3). Wavefunctions of HF 6‐31g** quality were calculated for every molecule and the molecular electric potential (MEP) was modeled with net atomic charges. Methylene bisector charges were used for all CH₂ and CH₃ groups, and ring center site charges were added if necessary to fit the MEP. For some structures, upward scaling of the MEP to simulate intermolecular polarization gave better results. MEP interaction generally gave a satisfactory representation of weak C HO hydrogen bonding. Medium strength C HO hydrogen bonding was modeled by reducing the repulsion of the involved hydrogen atoms. Crystal structures of several biologically interesting molecules were modeled with the force field, yielding good results. These molecules include aspirin, sucrose, β‐cellobiose, β‐lactose, progesterone, testosterone, prostaglandin E1, cholesterol acetate, and stearic acid. © 2000 John Wiley & Sons, Inc. J Comput Chem 22: 1–20, 2001     

Three dimensional model of severe acute respiratory syndrome coronavirus helicase ATPase catalytic domain and molecular design of severe acute respiratory syndrome coronavirus helicase inhibitors

The modeling of the severe acute respiratory syndrome coronavirus helicase ATPase catalytic domain was performed using the protein structure prediction Meta Server and the 3D Jury method for model selection, which resulted in the identification of 1JPR, 1UAA and 1W36 PDB structures as suitable templates for creating a full atom 3D model. This model was further utilized to design small molecules that are expected to block an ATPase catalytic pocket thus inhibit the enzymatic activity. Binding sites for various functional groups were identified in a series of molecular dynamics calculation. Their positions in the catalytic pocket were used as constraints in the Cambridge structural database search for molecules having the pharmacophores that interacted most strongly with the enzyme in a desired position. The subsequent MD simulations followed by calculations of binding energies of the designed molecules were compared to ATP identifying the most successful candidates, for likely inhibitors—molecules possessing two phosphonic acid moieties at distal ends of the molecule.     

Prediction of enzyme binding: human thrombin inhibition study by quantum chemical and artificial intelligence methods based on X-ray structures

Thrombin is a serine protease which plays important roles in the human body, the key one being the control of thrombus formation. The inhibition of thrombin has become a target for new antithrombotics. The aim of our work was to (i) construct a model which would enable us to predict Ki values for the binding of an inhibitor into the active site of thrombin based on a database of known X-ray structures of inhibitor-enzyme complexes and (ii) to identify the structural and electrostatic characteristics of inhibitor molecules crucially important to their effective binding. To retain as much of the 3D structural information of the bound inhibitor as possible, we implemented the quantum mechanical/molecular mechanical (QM/MM) procedure for calculating the molecular electrostatic potential (MEP) at the van der Waals surfaces of atoms in the protein's active site. The inhibitor was treated quantum mechanically, while the rest of the complex was treated by classical means. The obtained MEP values served as inputs into the counter-propagation artificial neural network (CP-ANN), and a genetic algorithm was subsequently used to search for the combination of atoms that predominantly influences the binding. The constructed CP-ANN model yielded Ki values predictions with a correlation coefficient of 0.96, with Ki values extended over 7 orders of magnitude. Our approach also shows the relative importance of the various amino acid residues present in the active site of the enzyme for inhibitor binding. The list of residues selected by our automatic procedure is in good correlation with the current consensus regarding the importance of certain crucial residues in thrombin's active site.     

药物设计中的三维结构搜索方法

利用分子的三维特征, 在三维结构数据库中进行三维结构搜索的方法是一种发现先导化合物的快捷而有效的方法, 已经得到了广泛的重视。本文综述了三维结构搜索方法的原理、发展及其近年来在药物分子设计中的应用。     

Similarity searching in databases of three-dimensional molecules and macromolecules.

This paper discusses algorithmic techniques for measuring the degree of similarity between pairs of three-dimensional (3-D) chemical molecules represented by interatomic distance matrices. A comparison of four methods for the calculation of 3-D structural similarity suggests that the most effective one is a procedure that identifies pairs of atoms, one from each of the molecules that are being compared, that lie at the center of geometrically-related volumes of 3-D space. This atom mapping method enables the calculation of a wide range of types of intermolecular similarity coefficient, including measures that are based on physicochemical data. Massively-parallel implementations of the method are discussed, using the AMT Distributed Array Processor, that achieve a substantial increase in performance when compared with a sequential implementation on a UNIX workstation. Current work involves the use of angular information and the extension of the method to field-based similarity searching. Similarity searching in 3-D macromolecules is effected by the use of a maximal common subgraph (MCS) isomorphism algorithm with a novel, graph-based representation of the tertiary structures of proteins. This algorithm is being used to identify similarities between the 3-D structures of proteins in the Brookhaven Protein Data Bank; its use is exemplified by searches involving the NAD-binding fold motif.     

The discovery of novel auxin transport inhibitors by molecular modeling and three-dimensional pattern analysis

Summary Molecular modeling techniques and three-dimensional (3D) pattern analysis have been used to investigate the chemical and steric properties of compounds that inhibit transport of the plant hormone auxin. These compounds bind to a specific site on the plant plasma membrane characterized by its affinity for the herbicide N-1-naphthylphthalamic acid (NPA). A 3D model was derived from critical features of a set of ligands for the NPA receptor, a suggested binding conformation is proposed, and implications for the topographical features of the NPA receptor are discussed. This model, along with 3D structural analysis techniques, was then used to search the Abbott corporate database of chemical structures. Of the 467 compounds that satisfied the criteria of the model, 77 representative molecules were evaluated for their ability to compete for the binding of [³H]NPA to corn microsomal membranes. Nineteen showed activity that ranged from 16 to 85% of the maximum NPA binding. Four of the most active of these, representing chemical classes not included in the original compound set, were also found to inhibit polar auxin transport through corn coleoptile sections. Thus, this study demonstrates that 3D analysis techniques can identify active, novel ligands for biochemical target sites with concomitant physiological activity.     

SPLICE: A program to assemble partial query solutions from three-dimensional database searches into novel ligands

Summary SPLICE is a program that processes partial query solutions retrieved from 3D, structural databases to generate novel, aggregate ligands. It is designed to interface with the database searching program FOUNDATION, which retrieves fragments containing any combination of a user-specified minimum number of matching query elements. SPLICE eliminates aspects of structures that are physically incapable of binding within the active site. Then, a systematic rule-based procedure is performed upon the remaining fragments to ensure receptor complementarity. All modifications are automated and remain transparent to the user. Ligands are then assembled by linking components into composite structures through overlapping bonds. As a control experiment, FOUNDATION and SPLICE were used to reconstruct a know HIV-1 protease inhibitor after it had been fragmented, reoriented, and added to a sham database of fifty different small molecules. To illustrate the capabilities of this program, a 3D search query containing the pharmacophoric elements of an aspartic proteinase-inhibitor crystal complex was searched using FOUNDATION against a subset of the Cambridge Structural Database. One hundred thirty-one compounds were retrieved, each containing any combination of at least four query elements. Compounds were automatically screened and edited for receptor complementarity. Numerous combinations of fragments were discovered that could be linked to form novel structures, containing a greater number of pharmacophoric elements than any single retrieved fragment.     

PharmID: pharmacophore identification using Gibbs sampling

The binding of a small molecule to a protein is inherently a 3D matching problem. As crystal structures are not available for most drug targets, there is a need to be able to infer from bioassay data the key binding features of small molecules and their disposition in space, the pharmacophore. Fingerprints of 3D features and a modification of Gibbs sampling to align a set of known flexible ligands, where all compounds are active, are used to discern possible pharmacophores. A clique detection method is used to map the features back onto the binding conformations. The complete algorithm is described in detail, and it is shown that the method can find common superimposition for several test data sets. The method reproduces answers very close to the crystal structure and literature pharmacophores in the examples presented. The basic algorithm is relatively fast and can easily deal with up to 100 compounds and tens of thousands of conformations. The algorithm is also able to handle multiple binding mode problems, which means it can superimpose molecules within the same data set according to two different sets of binding features. We demonstrate the successful use of this algorithm for multiple binding modes for a set of D2 and D4 ligands.