David Ryan Koes, Carlos J. Camacho, ZINCPharmer: pharmacophore search of the ZINC database, Nucleic Acids Analysis, Quantity 40, Subject W1, 1 July 2012, Pages W409–W414, https://doi.org/10.1093/nar/gks378
Summary
INTRODUCTION
A pharmacophore describes the structural association of the important molecular options of an interplay between a ligand and its receptor. Looking chemical libraries for compounds that match a particular pharmacophore is a longtime methodology of digital screening (1–3). The 2 predominant challenges of pharmacophore-based digital screening are figuring out a consultant pharmacophore for an interplay after which figuring out the compounds inside a related chemical library that match the pharmacophore. ZINCPharmer is a pharmacophore search engine for purchasable chemical area that addresses each these challenges.
An interplay pharmacophore could also be elucidated from a set of recognized lively ligands by figuring out a consensus pharmacophore that’s conformationally accessible to all these ligands (1,4). These methods don’t require a ligand-bound construction, however could also be computationally demanding if the enter set accommodates many versatile ligands. PharmaGist (5) is a free internet server that may establish a consensus pharmacophore of a set of as much as 32 ligands in a couple of minutes. Alternatively, structure-based approaches require a ligand-bound construction and establish a possible pharmacophore by analyzing the interplay website (6). ZINCPharmer gives a mechanism for deriving an preliminary pharmacophore speculation instantly from constructions throughout the PDB (Protein Knowledge Financial institution), and in addition helps importing pharmacophore definitions developed utilizing extra computationally demanding approaches carried out in third-party instruments.
Given a library of specific compound conformations, conformers that match a 3D pharmacophore could be discovered utilizing both fingerprint-based (7–9) or alignment-based (4,10) approaches. Fingerprints are effectively fitted to similarity metrics (11), however, since they discretize the pharmacophore illustration, present inexact outcomes. The EDULISS (12) on-line database gives fingerprint-based screening of a single-conformer library of some million compounds, however the question fingerprint have to be manually constructed from pairwise distance constraints. Alignment-based approaches produce extra correct and interpretable outcomes, on the expense of extra computation. For instance, a library of fewer than 1,000,000 conformers could take minutes or hours to display screen (13). Nevertheless, since there are considerably fewer protein targets than there are doable ligands, alignment-based pharmacophore screening can be utilized successfully when performing a reverse display screen that identifies matching protein targets as a substitute of ligands. PharmMapper (14) takes as enter a single ligand and screens a database of over 7000 receptors for potential targets.
Each fingerprint and alignment-based approaches usually consider each conformer within the library, leading to search occasions that scale with the dimensions of the database. Newer strategies, resembling Pharmer (15) and Recore (16) use indexing approaches in order that search occasions scale with the complexity and breadth of the question, not the dimensions of the library. ZINCPharmer makes use of the open-source Pharmer software program to allow the interactive search of greater than 176 million conformations in just some minutes, if not seconds.
METHODS
ZINCPharmer searches a database of conformations calculated from the purchasable compounds of the ZINC database (17). ZINC is a complete assortment of commercially out there, biologically related compounds appropriate for screening. Purchasable compounds have an anticipated availability of <10 weeks and are both out there from vendor inventory or are make-on-demand. The ZINCPharmer library is synchronized with the ZINC library on a month-to-month foundation. Compounds are each added and eliminated to keep up consistency and make sure that solely at present purchasable compounds are retained. ZINC compounds are transformed into 3D conformations utilizing omega2 from OpenEye Scientific Software program (http://eyesopen.com). Conformers are generated utilizing the default settings and -rms.7, which improves the sampling of conformational area in comparison with the default setting of .5 (18). The ten greatest conformers are saved. The generated conformers are transformed into an environment friendly search format utilizing the Pharmer (15) open-source software program. Pharmer identifies hydrophobic, hydrogen bond donor/acceptor, optimistic/detrimental ions and fragrant pharmacophore options utilizing the SMARTS matching performance of the OpenBabel toolkit (19). Presently, the default set of SMARTS definitions is used, however these are topic to refinement primarily based on consumer enter. These options are saved in an environment friendly spatial index to assist the fast search of enormous chemical libraries. For instance, the search proven in Determine 1 took lower than 3 seconds.
The graphical consumer interface (Determine 1) for outlining, refining and visualizing pharmacophore queries and their outcomes is carried out utilizing JavaScript and the Java-based Jmol (http://www.jmol.org/) molecular viewer. A contemporary, requirements compliant browser with a current Java plugin is required. Session state, which incorporates the pharmacophore definitions, could be saved in a human-readable JSON (JavaScript Object Notation) format and the aligned search outcomes could be saved within the sdf molecular format. An web discussion board hosts a consumer information and gives technical assist.DEFINING A PHARMACOPHORE QUERY
Utilizing the Pharmer software program, ZINCPharmer can mechanically extract a set of pharmacophore options from molecular construction. Every function consists of the function sort (hydrophobic, hydrogen bond donor/acceptor, optimistic/detrimental ion or fragrant), a place, and a search radius. Determine 2 illustrates the assorted strategies for creating an preliminary question. Options could also be derived from a single ligand construction, a protein–ligand construction, a protein–protein construction or from the output of third-party software program.
From ligand construction
Any single-conformer molecular construction file that’s appropriate with OpenBabel (19) could also be uploaded to outline a set of pharmacophore options. All recognized options of the molecule are enabled as pharmacophore question options. Nevertheless, since by itself a ligand gives no details about the character of an interplay, the outcome is just not a real pharmacophore. For example, though low-energy conformers are sometimes shut in configuration to the certain constructions (18), with out extra data it’s inconceivable to separate interacting options from non-interacting options. As an alternative, the pharmacophore derived from a single ligand construction must be considered a 3D similarity search.
If the receptor construction is understood, a versatile docking of the ligand can generate a customized protein–ligand construction from which ZINCPharmer can mechanically derive an interplay pharmacophore. Alternatively, if there are various recognized binders then a consensus pharmacophore could be elucidated (1,4) utilizing software program resembling Chemical Computing Group’s MOE (http://www.chemcomp.com/), Inte:Ligand’s LigandScout (http://www.inteligand.com/), or PharmaGist (5) and the outcome could be imported into ZINCPharmer.
From protein-ligand construction
When supplied with each a receptor and bound-ligand construction, ZINCPharmer will mechanically establish an interplay pharmacophore. All doable pharmacophore options on the ligand are computed, however solely these which are inside a distance cutoff of complimentary options on the receptor are enabled. Hydrogen bond acceptors/donors have to be inside 4 Å of a hydrogen bond donor/acceptor on the receptor. Charged options have to be inside 5 Å of an oppositely charged function on the receptor. Fragrant function have to be inside 5 Å of a receptor fragrant function. A ligand hydrophobic function have to be inside 6 Å of at the very least three hydrophobic options on the receptor in an effort to require a point of buriedness. The space cutoffs are meant to be permissive and no angular cutoffs are utilized since it’s conceptually simpler for a consumer to cut back the variety of options in a pharmacophore question than to extend them (which requires investigating a a lot bigger variety of potential options).
If the protein–ligand construction exists within the PDB, then a shortcut is obtainable on the ZINCPharmer dwelling web page the place the consumer want solely enter the PDB accession code, choose the specified ligand and click on the Begin button (Determine 2). The corresponding ligand and receptor constructions in addition to their interplay pharmacophore will mechanically be loaded into a brand new ZINCPharmer session.
For customized protein–ligand constructions, for instance, the results of a docking examine, the receptor and ligand have to be uploaded individually. With the intention to establish the interplay pharmacophore, the receptor have to be uploaded first.
From protein–protein interplay construction
ZINCPharmer is built-in with PocketQuery (http://pocketquery.csb.pitt.edu), an internet site that identifies protein–protein interplay (PPI) inhibitor beginning factors from PPI construction. Utilizing a consensus scoring scheme (20), PocketQuery identifies a small set of interacting residues in a PPI construction whose mimicry by a small molecule is prone to inhibit the interplay. Inside the PocketQuery interface, as proven in Determine 2, the chosen set of residues could be exported on to ZINCPharmer. The interplay pharmacophore between these ligand residues and the receptor will than be mechanically generated as with a protein–ligand construction.
From third social gathering software program
ZINCPharmer contains assist for importing pharmacophore definitions represented in both PH4 format, utilized by MOE, or PML format, utilized by LigandScout. Moreover, the specialised mol2 format exported by PharmaGist (5) is acknowledged as a hybrid pharmacophore definition and ligand construction file. These packages can be utilized to elucidate a consensus pharmacophore from a set of lively compounds. ZINCPharmer can then import the outcome and rapidly establish all matching hits. Nevertheless, there are a number of variations between the pharmacophore recognition routines and alignment insurance policies of various software program packages (21). Particularly, the identification and positioning of hydrophobic options has essentially the most variation between software program packages. Consequently, ZINCPharmer searches utilizing an externally outlined pharmacophore will lead to an overlapping, however not equivalent, set of hits in contrast with a search carried out utilizing the software program that generated the pharmacophore.
REFINING A QUERY – “zinc database”
Though ZINCPharmer is able to mechanically extracting a pharmacophore from an interplay, it’s anticipated that the consumer will additional refine the question to boost its specificity and applicability. This may be finished by modifying the properties of the question or by making use of filters to the outcomes.
Question editor
Each pharmacophore function is a row within the question editor and has a pharmacophore class (hydrophobic, hydrogen bond donor/acceptor, optimistic/detrimental ion or fragrant), a place laid out in Cartesian coordinates, a radius representing the tolerance sphere to go looking this place and an enabled/disabled setting. The pharmacophore question editor, proven within the backside left of Determine 1, helps the interactive modifying of those options, that are proven as spheres within the molecular viewer as seen within the high left of Determine 1. Options could also be chosen both within the question editor desk or instantly within the molecular viewer by clicking on the related sphere. Chosen options could also be batch processed (enabled, disabled, deleted or duplicated) by a contextual menu accessible by right-clicking the chosen rows.
Some options have extra choices distinctive to their pharmacophore class which are accessible by a drop down menu. Hydrogen bond donors/acceptors have an elective directionality, as proven within the drop down menu of Determine 1. The question vector is matched towards a precomputed vector on the ligand. Because the precise route of the hydrogen bond is particular to the geometry of the interface, this match is essentially approximate, and due to this fact a big tolerance in angular deviation is carried out by default.
Fragrant options even have an elective directionality constraint that matches towards the conventional vector of the fragrant ring. Hydrophobic options have an elective constraint for specifying the variety of atoms collaborating within the hydrophobic space. For instance, if a small hydrophobe, resembling a methyl group, is desired, then the utmost variety of atoms could be constrained to at least one. Alternatively, if a big, space-filling group is desired, resembling an aliphatic ring, the minimal variety of atoms could be constrained to 5 or increased.
Filters
The outcomes could be filtered each when it comes to the variety of returned outcomes and the properties of the returned outcomes. The variety of hits could be lowered by specifying a restrict on the variety of completely different orientations returned for every conformation (‘Max Hits per Conf’), the variety of completely different orientations of various conformations returned for every molecule (‘Max Hits per Mol’), or the whole variety of hits returned (‘Max Total Hits’). In all instances, the search is terminated as quickly because the restrict is reached with no assure that the returned hits have the very best root imply squared deviation (RMSD) to the question.
Every orientation of a conformer outcomes from a unique mapping and alignment of pharmacophore options on the ligand to the question options. If the question has many levels of symmetry or tightly spaced options, decreasing the variety of orientations returned could considerably scale back the variety of hits that have to be analyzed with out omitting important positional variations. Decreasing the variety of hits per a molecule is especially helpful when solely the 2D properties of the outcomes will probably be analyzed and solely a single consultant of every molecule is required. Decreasing the whole variety of hits is useful when the post-screening evaluation is computationally intensive and solely a sampling of the outcomes is required.
The outcomes listing can be filtered by most RMSD. The orientation of the hits is computed utilizing a weighted RMSD calculation (15), however the reported worth is the unweighted RMSD between the calculated orientation and the question. Filtering by RMSD restricts the hits to those who have the perfect general geometric match to the question. Moreover, hits could be filtered by the molecular properties of molecular weight (in Daltons) and variety of rotatable bonds, each of which have been implicated as helpful properties for figuring out ‘drug-like’ molecules (22).
PHARMACOPHORE SEARCH
Having outlined a pharmacophore, looking for matching purchasable compounds is so simple as clicking the ‘Submit Query’ button. Searches take wherever from a couple of seconds to some minutes. Queries with extra options, queries with many hydrophobic options (that are the most typical options), queries with massive search tolerances and symmetric queries (which require the processing of many orientations per an identical conformer) can have longer search occasions. Outcomes are returned and displayed within the outcomes browser as they’re discovered. An orientation of a conformer is just returned as a success if all of the matching options are throughout the specified search tolerances of the question when the conformer is aligned to reduce the weighted RMSD.
RESULTS VISUALIZATION
The outcomes of a search are displayed within the outcomes browser proven in Determine 1. Every hit represents a novel orientation of a conformation to the question. For every hit, the ZINC identifier, RMSD to the question, molecular weight (‘Mass’), and variety of rotatable bonds (‘RBnds’) is proven. The ZINC identifier is a hyperlink that factors to the corresponding compound internet web page within the ZINC database the place buying data could also be discovered. The outcomes could also be sorted by any of the numerical properties by clicking on the property heading within the outcomes desk. The entire set of oriented hits could also be saved to an sdf file by the ‘Save Results’ button. The hits on this file are unordered and embody the RMSD to the question as further information connected to every molecule. This file is instantly helpful as enter to a secondary screening protocol resembling rating by vitality minimization.
Particular person hits are visualized with the question and a receptor (if current) by clicking on the corresponding row within the outcomes browser. The viewer tab accommodates a large assortment of colours and kinds (wireframe, stick, spheres, and so on.) for visualizing the outcomes, the question ligand, the receptor residues and the receptor floor.
“zinc database”