In-silico design tool & External Computations

In-silico design tool & External Computations

This functionality enables the possibility to draw and/or import a molecule and to perform predictions using MetaSite, Moka, Descriptor calculation or ADME prediction. The user can then decide if the structure should be imported to the database and in case that it is case the system will check if the structure is exiting then it will create the computation if needed, and in the opposite case then it will generate a new compound. 

Compound Library – In-silico designer 

MetaSite

This option enables the user to compute the Site of Metabolism for phase I metabolism. To be able to use this function the user would have to set a MetaSite valid license in the settings of the program. 

MetaSite is a computational procedure to predict metabolism issues related to cytochrome-mediated reactions in phase I metabolism. Validation of the method on real company data has shown that the primary site of metabolism was found in the top three MetaSite predictions for more than 85% of the cases. Validation on literature data gave even better results. Moreover, MetaSite provides the structure of the metabolites formed with a ranking derived from the site of metabolism predictions. 

In addition to predicting the site of metabolism, the method also highlights the atoms in the molecule that contribute to the prediction (i.e., they help direct the molecule in the cytochrome cavity such that the site of metabolism is in proximity to the catalytic center). Directly blocking the primary site of metabolism can risk creating an inhibitor of the cytochrome or may negatively affect the activity or selectivity of the compound towards its therapeutic target. Modifying these contributing regions that most influence the site of metabolism can bypass both potential problems. 

The MetaSite procedure is completely automated and does not require any user assistance. All the work can be handled and submitted in a batch queue.  

The basic concept of MetaSite is to compare the interaction patterns computed inside the protein with the 3D structure of the ligand. The information obtained from a user-friendly interface can be easily translated into decisions in the drug discovery process. 

MetaSite suggests positions on ligand molecules which should be modified to avoid metabolic degradation or to promote the release of an active compound from a pro-drug molecule. 

Superposition of molecules, training set preparation, ab initio computation, or statistical analyses are not needed. MetaSite runs both on workstations and personal computers and supports a large variety of operating systems. 

MetaSite is specifically designed for: 

• In silico Metabolism predictions. 
• Prediction of the site of metabolism for different cytochromes. 
• Prediction of the atoms that contribute to the site of metabolism. 
• Metabolite identification. 
• Comparison of the site of metabolism due to different cytochromes.  
• Identification of cytochrome isoform(s) responsible for the metabolism of substrates.  
• Considering chemical reactivity towards the oxidation. 

MetaSite can be used for: 

• Automatic prediction of the site of metabolism for xenobiotics. 
• Metabolite identification. 
• Relative metabolite retention time prediction. 
• Lead optimization: suggesting the place in the molecule to be protected against metabolic degradation.  
• Lead optimization; suggesting the site for chemical modification in pro-drug strategy.  
• Lead optimization: suggesting the part of the molecule that can potentially produce mechanism-based inhibition.  
• Working with large batches of molecules in a single run. 
• Generating useful quantitative description for the protein with reference to the reactive center. 
• Patent protection. 

MetaSite is not a training set dependent method, a docking algorithm, or a force-field method, although it is linked to the GRID methodology used to generate the 3D input Molecular Interaction Field maps for the protein description. 

Furthermore, MetaSite is not a conformational analysis tool although it does use a minimization tool to produce conformers. MetaSite is not a statistical tool nor is it a semi-empirical or ab-initio method, or an QSAR or 3D-QSAR method either, or does not use QSAR models for the Site of Metabolism prediction (so it is not training set dependent). 

Unique features already present in MetaSite: 

• FMO enzymes (FMOs) play a key role in detoxification and/or bio-activation of specific pharmaceuticals and xenobiotics bearing nucleophilic centers. However, in silico methods for FMO metabolism prediction were not yet available. This program reports, for the first time, a substrate-specificity and catalytic-activity model for FMO3, the most relevant isoform of the FMOs in humans. The model has also been particularly useful to design compounds with optimal clearance. 
• When other factors are constant, a nucleophilicity scale can be used to estimate the likelihood that a chemical group suffers an electrophilic attack by FMO enzymes. In MetaSite nucleophilicity in water is now fast estimated by the charge, polarizability, solvation, and HOMO energy of a chemical moiety.  
• Most SoM-prediction software use very sophisticated approaches to compute chemical reactivity. However, proximity effects between the Heme-protein complex and substrates are not considered; the pKa of protein amino acids and substrates chemical moieties not considered. Conversely MetaSite uses MoKa software (embedded in the procedure) to automatically determine the proper ionization state of substrates and protein residues upon binding. This provides a more realistic binding simulation, with a net effect on SoM and quantification of the major reactions occurring.  
• Aldehyde oxidase (AOX) is a molibdo-flavoenzyme that raised great interest in recent years, since its contribution in xenobiotics metabolism has not always been found before clinical trials, with resulting negative effects on the fate of new potential drugs. The fundamental role of AOX in metabolizing xenobiotics is also due to the attempt of medicinal chemists to stabilize candidates toward cytochrome P450 activity, which increases the risk for new compounds to be susceptible to AOX nucleophile attack. Therefore, novel strategies to predict the potential liability of new entities towards AOX enzyme are urgently needed to increase effectiveness, reduce costs, and prioritize experimental studies. In the present software implementation, we present the most up-to-date computational method to predict liability towards human AOX, for applications in drug design and pharmacokinetic optimization. The method was developed using a large dataset of homogeneous experimental data, which is also showed as supplementary material.  
• The model combines the knowledge acquired from both experimental and in silico sources. Indeed, more than 600 compounds (acquired or synthesized to meet the needs of assessing structure-activity relationships) were experimentally tested in human liver cytosol (HLC) to generate the largest database of homogeneous experimental data publicly available so far. Applied in an early phase, the method should facilitate a good understanding of the passivity of AOX-catalyzed metabolism and suggest potential modifications to increase the stability of a xenobiotics to AOX-catalyzed reactions.  
• The function describing the recognition between the specific CYP (Cytochrome P450) protein and the substrate when the substrate is positioned in the CYP protein and exposes the atom i toward the heme (accessibility, called Ei), has been substituted with a more complex and precise function based on the FLAP algorithm. The score is not calculated by evaluating the thermodynamics of the enzyme-substrate interaction but considering the three-dimensional complementarity between the fields of the protein and those induced by the substrate. The calculation of the shape of the protein cavity has been improved and the exposition function created by this new method is more selective, that is, it is even better at discriminating similar exposures. The substrate reactivity computed from molecular orbital calculations has also been improved since the fragments that will eventually recompose the substrate will not have a calculated Ri typical of the isolated fragments, but their Ri will be ‘evolved’ to reflect the global structure of the substrate.  
• The two factors that contribute to the creation of the P-SoM function in the original equation (recognition and reactivity) were weighted equally. However, their respective contributions very much depend on both the cytochrome being used and on the substrate molecule. It would be more correct to use weight coefficients that are based on the type of the enzyme cavity and the structure of the substrate. This feature in MetaSite is called reactivity-equalization. The use of this function allows the automatic recalculation of the weights in the equation to consider the enzyme cavity (which is determined by the type of CYP being used) and the interactions of the substrate with the amino acids within this cavity. One result of the reactivity-equalization function is that the weight constant of the reactivity component of the equation will be increased for small to medium substrates in large enzymes (3A4) and decreased for medium to large molecules in enzymes with small cavities.  
• An automatic bio isosteric replacement of fragments called MetaDesign is provided with the intent to help design molecules with improved metabolic properties (in general to reduce metabolic liability). Finally, the molecular moieties that influence the exposure the most are more precisely computed and nicely reported in a new dedicated graphical interface called 32D visualization.  

MoKa

MoKa is an application for fast and accurate prediction of pKa, logP, and logD of organic compounds. In medicinal chemistry these properties are particularly important because they affect ADMET and activity of drug candidates. In this chapter, you will find a step-by-step guide on how to use the graphical user interface and the command line tool of MoKa. Every function accessible from the graphical interface is also accessible from the command line.  

Descriptor calculation

The descriptors are computed using the VolSurf + program. VolSurf+ is an advanced computational procedure aimed to produce and to explore the Physico-chemical property space of a molecule (or library of molecules) starting from 3D maps of interaction energies between the molecule and chemical probes (GRID based Molecular Interaction Fields, or MIFs). VolSurf+ compresses the information present in 3D maps into numerical descriptors optimized for ADME models and virtual screening which are simple to understand and to interpret. Superposition of molecules is not needed, since VolSurf+ descriptors are superposition independent. 3D structure generation is not needed, since VolSurf+ automatically generates the 3D (minimized) structure of molecules. Conformational analysis is not needed, since VolSurf+ automatically generates up to 50 conformers for each structure under study. A short comparison between VolSurf+ and VolSurf is reported in Chapter 2. 

VolSurf+ is specifically designed for:  

• In silico DMPK (Drug Metabolism and Pharmacokinetics) predictions 
• ADME database analyses and determination of filters for early phase drug discovery 
• DMPK properties calculation to bias combinatorial libraries 
• Pattern recognition techniques for specific ADME properties 
• Fast generation of quantitative and lattice independent molecular descriptors for Quantitative Structure-Property Relationships 
• Fast quantification on distribution, as well as magnitude of molecular surface polarity 
• Extraction of chemical information from 3D molecular image maps 
• Obtaining useful descriptors for optimizing pharmacokinetic properties in lead optimization 
• Working with small, medium, and large molecules, as well as with biopolymers (DNA and peptides). 

VolSurf+ can be used for automated generation of molecular descriptors designed for the computation of ADME properties 

VolSurf+ descriptors and features:  

• Molecular size and molecular shape descriptors 
• Hydrophilic and hydrophobic regions quantification 
• Integy moments and Capacity factors 
• Amphiphilic moment, Hydro-Lipo balance 
• Molecular diffusivity, LogP, LogD, pH-dependent solubility, molecular flexibility 
• 3D Pharmacophoric descriptors 

A three-dimensional map (3D map) may be viewed as a 3D matrix which contains attractive and repulsive forces between a chemical probe and a target molecule. A 3D map is an image of the target-probe molecular interaction, in which each pixel contains information about the Cartesian coordinates x,y,z and a chemical property.  

The amount of information contained in a 3D map is related to the interacting molecular partners. Sometimes visual inspection is not sufficient to extract useful information since a large amount of information is coded and hidden in the sign and magnitude of the grid node forces, in the position of the grid nodes, in the relationships between grid nodes and in other functional relationships. 

Although 3D-QSAR models can be obtained from these 3D maps (CoMFA [6] or GOLPE [7] procedures), the usefulness of the models is limited by their difficult interpretation. Further problems arise because of alignment and molecular flexibility. 

Specialized tools are needed to facilitate the extraction of useful descriptors from 3D Molecular Interaction Field images and to link experimental observations with molecular structures. 

VolSurf+ is a computational procedure designed to produce and to explore the Physico-chemical property space of a molecule starting from 3D interaction energy maps. The basic concept of VolSurf+ is to compress the information present in 3D maps into a few quantitative numerical descriptors that are remarkably simple to understand and to interpret. 

Compression of information is made using image analysis software. Each 3D map is considered as 3D image, but the image compression process is made adding chemical knowledge. VolSurf+ does this by selecting the most proper descriptors and parametrization according to the type of the 3D map under study. 

Interaction fields with a water probe (OH2), a hydrophobic probe (DRY) plus an H-bond donor (NH) and an H-bond acceptor (=O) probes are calculated all around the target molecules as in the program GRID.  

VolSurf+ has the nice advantage of producing descriptors using the 3D information embedded in any map. Not all the information can be transferred from 3D to 2D descriptors, but practical examples do exist showing that relevant information is extracted. Moreover, the VolSurf+ transformation is easy to understand, fast to compute, the descriptors have a clear chemical meaning and are lattice independent, and some of them can be projected back into the original 3D map from which they were obtained. VolSurf+ descriptors can be obtained for small, medium, and large molecules, provided the input file is one of those supported by the procedure.  

ADME prediction

The ADME Prediction enable the calculation of the ADME parameters by the VolSurf+ application.  

VolSurf+ is an advanced computational procedure aimed to produce and to explore the Physico-chemical property space of a molecule (or library of molecules) starting from 3D maps of interaction energies between the molecule and chemical probes (GRID based Molecular Interaction Fields, or MIFs (Molecular Interaction Fields)). VolSurf+ compresses the information present in 3D maps into numerical descriptors optimized for ADME models and virtual screening which are simple to understand and to interpret. Superposition of molecules is not needed, since VolSurf+ descriptors are superposition independent. 3D structure generation is not needed, since VolSurf+ automatically generates the 3D (minimized) structure of molecules. Conformational analysis is not needed, since VolSurf+ automatically generates up to 50 conformers for each structure under study. A short comparison between VolSurf+ and VolSurf is reported in Chapter 2. 

VolSurf+ is specifically designed for:  

• In silico DMPK predictions 
• ADME database analyses and determination of filters for early phase drug discovery 
• DMPK properties calculation to bias combinatorial libraries 
• Pattern recognition techniques for specific ADME properties 
• Fast generation of quantitative and lattice independent molecular descriptors for Quantitative Structure-Property Relationships 
• Fast quantification on distribution, as well as magnitude of molecular surface polarity 
• Extraction of chemical information from 3D molecular image maps 
• Obtaining useful descriptors for optimizing pharmacokinetic properties in lead optimization 
• Working with small, medium, and large molecules, as well as with biopolymers (DNA and peptides). 

VolSurf+ can be used for automated generation of molecular descriptors designed for the calculation of ADME properties 

• VolSurf+ descriptors and features:  
• Molecular size and molecular shape descriptors 
• Hydrophilic and hydrophobic regions quantification 
• Integy moments and Capacity factors 
• Amphiphilic moment, Hydro-Lipo balance 
• Molecular diffusivity, LogP, LogD, pH-dependent solubility, molecular flexibility 
• 3D Pharmacophoric descriptors 

A three-dimensional map (3D map) may be viewed as a 3D matrix which contains attractive and repulsive forces between a chemical probe and a target molecule. A 3D map is an image of the target-probe molecular interaction, in which each pixel contains information about the cartesian coordinates x,y,z and a chemical property.  

The amount of information contained in a 3D map is related to the interacting molecular partners. Sometimes visual inspection is not sufficient to extract useful information since a large amount of information is coded and hidden in the sign and magnitude of the grid node forces, in the position of the grid nodes, in the relationships between grid nodes and in other functional relationships. 

Although 3D-QSAR models can be obtained from these 3D maps (CoMFA [6] or GOLPE [7] procedures), the usefulness of the models is limited by their difficult interpretation. Further problems arise because of alignment and molecular flexibility. 

Specialized tools are needed to facilitate the extraction of useful descriptors from 3D Molecular Interaction Field images and to link experimental observations with molecular structures. 

VolSurf+ is a computational procedure designed to produce and to explore the physicochemical property space of a molecule starting from 3D interaction energy maps. The basic concept of VolSurf+ is to compress the information present in 3D maps into a few quantitative numerical descriptors that are quite simple to understand and to interpret. 

Compression of information is made using image analysis software. Each 3D map is considered as 3D image, but the image compression process is made adding chemical knowledge. VolSurf+ does this by selecting the most proper descriptors and parameterization according to the type of the 3D map under study. 

Interaction fields with a water probe (OH2), a hydrophobic probe (DRY) plus an H-bond donor (NH) and an H-bond acceptor (=O) probes are calculated all around the target molecules as in the program GRID.  

VolSurf+ has the nice advantage of producing descriptors using the 3D information embedded in any map. Not all the information can be transferred from 3D to 2D descriptors, but practical examples do exist showing that relevant information is extracted. Moreover, the VolSurf+ transformation is easy to understand, fast to compute, the descriptors have a clear chemical meaning and are lattice independent, and some of them can be projected back into the original 3D map from which they were obtained. VolSurf+ descriptors can be obtained for small, medium, and large molecules, provided the input file is one of those supported by the procedure.