scholarly journals Teaching a Neural Network to Attach and Detach Electrons from Molecules

2020 ◽  
Author(s):  
Roman Zubatyuk ◽  
Justin Smith ◽  
Benjamin T. Nebgen ◽  
Sergei Tretiak ◽  
Olexandr Isayev
Keyword(s):  
Density Functional ◽  
Organic Molecules ◽  
Chemical Elements ◽  
Chemical Reactivity ◽  
Machine Learning Algorithms ◽  
Training Data ◽  
Electrophilic Aromatic Substitution ◽  
Interatomic Potentials ◽  
Conceptual Density Functional Theory ◽  
Fukui Functions

<p>Physics-inspired Artificial Intelligence (AI) is at the forefront of methods development in molecular modeling and computational chemistry. In particular, interatomic potentials derived with Machine Learning algorithms such as Deep Neural Networks (DNNs), achieve the accuracy of high-fidelity quantum mechanical (QM) methods in areas traditionally dominated by empirical force fields and allow performing massive simulations. The applicability domain of DNN potentials is usually limited by the type of training data. As such, transferable models are aimed to be extensible in the description of chemical and conformational diversity of organic molecules. However, most DNN potentials, such as the AIMNet model we proposed previously, were parametrized for neutral molecules or closed-shell ions due to architectural limitations. In this work, we extend our AIMNet framework toward open-shell anions and cations. This model explores a new dimension of transferability by adding the charge-spin space. The resulting AIMNet model is capable of reproducing reference QM energies for cations, neutrals and anions with errors of 4.1, 2.1, 2.8 kcal/mol, respectively, compared to the reference QM simulations. The spin-charges have errors 0.01-0.06 electrons for small organic molecules containing nine chemical elements {H, C, N, O, F, Si, P, S and Cl}. Thus the proposed AIMNet model allows researchers to fully bypass QM calculations and derive the ionization potential, electron affinity, and conceptual Density Functional Theory quantities like electronegativity, hardness, and condensed Fukui functions. We show that these descriptors, along with learned atomic representations, could be used to model chemical reactivity through an example of regionselectivity in electrophilic aromatic substitution reactions.</p>

scholarly journals Teaching a Neural Network to Attach and Detach Electrons from Molecules

2020 ◽  
Author(s):  
Roman Zubatyuk ◽  
Justin Smith ◽  
Benjamin T. Nebgen ◽  
Sergei Tretiak ◽  
Olexandr Isayev
Keyword(s):  
Density Functional ◽  
Organic Molecules ◽  
Chemical Elements ◽  
Chemical Reactivity ◽  
Machine Learning Algorithms ◽  
Training Data ◽  
Electrophilic Aromatic Substitution ◽  
Interatomic Potentials ◽  
Conceptual Density Functional Theory ◽  
Fukui Functions

<p>Physics-inspired Artificial Intelligence (AI) is at the forefront of methods development in molecular modeling and computational chemistry. In particular, interatomic potentials derived with Machine Learning algorithms such as Deep Neural Networks (DNNs), achieve the accuracy of high-fidelity quantum mechanical (QM) methods in areas traditionally dominated by empirical force fields and allow performing massive simulations. The applicability domain of DNN potentials is usually limited by the type of training data. As such, transferable models are aimed to be extensible in the description of chemical and conformational diversity of organic molecules. However, most DNN potentials, such as the AIMNet model we proposed previously, were parametrized for neutral molecules or closed-shell ions due to architectural limitations. In this work, we extend our AIMNet framework toward open-shell anions and cations. This model explores a new dimension of transferability by adding the charge-spin space. The resulting AIMNet model is capable of reproducing reference QM energies for cations, neutrals and anions with errors of 4.1, 2.1, 2.8 kcal/mol, respectively, compared to the reference QM simulations. The spin-charges have errors 0.01-0.06 electrons for small organic molecules containing nine chemical elements {H, C, N, O, F, Si, P, S and Cl}. Thus the proposed AIMNet model allows researchers to fully bypass QM calculations and derive the ionization potential, electron affinity, and conceptual Density Functional Theory quantities like electronegativity, hardness, and condensed Fukui functions. We show that these descriptors, along with learned atomic representations, could be used to model chemical reactivity through an example of regionselectivity in electrophilic aromatic substitution reactions.</p>

scholarly journals Teaching a Neural Network to Attach and Detach Electrons from Molecules

2021 ◽  
Author(s):  
Roman Zubatyuk ◽  
Justin Smith ◽  
Benjamin T. Nebgen ◽  
Sergei Tretiak ◽  
Olexandr Isayev
Keyword(s):  
Machine Learning ◽  
Density Functional ◽  
Organic Molecules ◽  
Chemical Elements ◽  
Chemical Reactivity ◽  
Machine Learning Algorithms ◽  
Training Data ◽  
Electrophilic Aromatic Substitution ◽  
Interatomic Potentials ◽  
Conceptual Density Functional Theory

<p></p><p>Physics-inspired Artificial Intelligence (AI) is at the forefront of methods development in molecular modeling and computational chemistry. In particular, interatomic potentials derived with Machine Learning algorithms such as Deep Neural Networks (DNNs), achieve the accuracy of high-fidelity quantum mechanical (QM) methods in areas traditionally dominated by empirical force fields and allow performing massive simulations. The applicability domain of DNN potentials is usually limited by the type of training data. As such, transferable models are aimed to be extensible in the description of chemical and conformational diversity of organic molecules. However, most DNN potentials, such as the AIMNet model we proposed previously, were parametrized for neutral molecules or closed-shell ions due to architectural limitations. In this work, we extend machine learning framework toward open-shell anions and cations. We introduce AIMNet-NSE (Neural Spin Equilibration) architecture, which being properly trained, could predict atomic and molecular properties for an arbitrary combination of molecular charge and spin multiplicity. This model explores a new dimension of transferability by adding the charge-spin space. The AIMNet-NSE model is capable of reproducing reference QM energies for cations, neutrals, and anions with errors of about 2-3 kcal/mol, compared to the reference QM simulations. The spin-charges have errors ~0.01 electrons for small organic molecules containing nine chemical elements {H, C, N, O, F, Si, P, S and Cl}. <a>The AIMNet-NSE model allows to fully bypass QM calculations and derive the ionization potential, electron affinity, and conceptual Density Functional Theory quantities like electronegativity, hardness, and condensed Fukui functions with a speed up to 10<sup>4</sup> molecules per second on a single modern GPU.</a> We show that these descriptors, along with learned atomic representations, could be used to model chemical reactivity through an example of regioselectivity in electrophilic aromatic substitution reactions.</p><p></p>

scholarly journals Teaching a neural network to attach and detach electrons from molecules

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
Roman Zubatyuk ◽  
Justin S. Smith ◽  
Benjamin T. Nebgen ◽  
Sergei Tretiak ◽  
Olexandr Isayev
Keyword(s):  
Machine Learning ◽  
Density Functional ◽  
Chemical Reactivity ◽  
Closed Shell ◽  
Machine Learning Algorithms ◽  
Electrophilic Aromatic Substitution ◽  
Interatomic Potentials ◽  
Aromatic Substitution ◽  
Conceptual Density Functional Theory ◽  
Fukui Functions

AbstractInteratomic potentials derived with Machine Learning algorithms such as Deep-Neural Networks (DNNs), achieve the accuracy of high-fidelity quantum mechanical (QM) methods in areas traditionally dominated by empirical force fields and allow performing massive simulations. Most DNN potentials were parametrized for neutral molecules or closed-shell ions due to architectural limitations. In this work, we propose an improved machine learning framework for simulating open-shell anions and cations. We introduce the AIMNet-NSE (Neural Spin Equilibration) architecture, which can predict molecular energies for an arbitrary combination of molecular charge and spin multiplicity with errors of about 2–3 kcal/mol and spin-charges with error errors ~0.01e for small and medium-sized organic molecules, compared to the reference QM simulations. The AIMNet-NSE model allows to fully bypass QM calculations and derive the ionization potential, electron affinity, and conceptual Density Functional Theory quantities like electronegativity, hardness, and condensed Fukui functions. We show that these descriptors, along with learned atomic representations, could be used to model chemical reactivity through an example of regioselectivity in electrophilic aromatic substitution reactions.

scholarly journals Conceptual DFT as a Novel Chemoinformatics Tool for Studying the Chemical Reactivity Properties of the Amatoxin Family of Fungal Peptides

2019 ◽  
Vol 17 (1) ◽  
pp. 1133-1139 ◽  
Author(s):  
Norma Flores-Holguín ◽  
Juan Frau ◽  
Daniel Glossman-Mitnik
Keyword(s):  
Density Functional ◽  
Chemical Reactivity ◽  
Active Regions ◽  
Chemical Hardness ◽  
Conceptual Density Functional Theory ◽  
Fukui Functions ◽  
Reactivity Descriptors ◽  
Molecular Systems ◽  
Chemical Structures ◽  
Simple Protocol

AbstractThe chemical structures and molecular reactivities of the Amatoxin group of fungi-derived peptides have been determined by means of the consideration of a model chemistry that has been previously validated as well-behaved for our purposes. The reactivity descriptors were calculated on the basis of a methodological framework built around the concepts that are the outcome of the so called Conceptual Density Functional Theory (CDFT). This procedure in connection with the different Fukui functions allowed to identify the chemically active regions within the molecules. By considering a simple protocol designed by our research group for the estimation of the pKa of peptides through the information coming from the chemical hardness, these property has been established for the different molecular systems explored in this research. The information reported through this work could be of interest for medicinal chemistry researchers in using this knowledge for the design of new medicines based on the studied peptides or as a help for the understanding of the toxicity mechanisms exerted by them.

scholarly journals Chemical-Reactivity Properties, Drug Likeness, and Bioactivity Scores of Seragamides A–F Anticancer Marine Peptides: Conceptual Density Functional Theory Viewpoint

Computation ◽  
2019 ◽  
Vol 7 (3) ◽  
pp. 52 ◽  
Author(s):  
Norma Flores-Holguín ◽  
Juan Frau ◽  
Daniel Glossman-Mitnik
Keyword(s):  
Density Functional Theory ◽  
Active Sites ◽  
Density Functional ◽  
Chemical Reactivity ◽  
Drug Repurposing ◽  
Chemical Hardness ◽  
Conceptual Density Functional Theory ◽  
Functional Theory ◽  
Fukui Functions ◽  
Reactivity Descriptors

A methodology based on concepts that arose from Density Functional Theory (CDFT) was chosen for the calculation of global and local reactivity descriptors of the Seragamide family of marine anticancer peptides. Determination of active sites for the molecules was achieved by resorting to some descriptors within Molecular Electron Density Theory (MEDT) such as Fukui functions. The pKas of the six studied peptides were established using a proposed relationship between this property and calculated chemical hardness. The drug likenesses and bioactivity properties of the peptides considered in this study were obtained by resorting to a homology model by comparison with the bioactivity of related molecules in their interaction with different receptors. With the object of analyzing the concept of drug repurposing, a study of potential AGE-inhibition abilities of Seragamides peptides was pursued by comparison with well-known drugs that are already available as pharmaceuticals.

scholarly journals DFT investigations on arylsulphonyl pyrazole derivatives as potential ligands of selected kinases

2020 ◽  
Vol 18 (1) ◽  
pp. 857-873
Author(s):  
Kornelia Czaja ◽  
Jacek Kujawski ◽  
Radosław Kujawski ◽  
Marek K. Bernard
Keyword(s):  
Free Energy ◽  
Molecular Orbital ◽  
Density Functional ◽  
Chemical Reactivity ◽  
Chemical Hardness ◽  
Free Energy Of Solvation ◽  
Fukui Functions ◽  
Reactivity Descriptors ◽  
Condensed Fukui Functions ◽  
Charge Partitioning

AbstractUsing the density functional theory (DFT) formalism, we have investigated the properties of some arylsulphonyl indazole derivatives that we studied previously for their biological activity and susceptibility to interactions of azoles. This study includes the following physicochemical properties of these derivatives: electronegativity and polarisability (Mulliken charges, adjusted charge partitioning, and iterative-adjusted charge partitioning approaches); free energy of solvation (solvation model based on density model and M062X functional); highest occupied molecular orbital (HOMO)–lowest occupied molecular orbital (LUMO) gap together with the corresponding condensed Fukui functions, time-dependent DFT along with the UV spectra simulations using B3LYP, CAM-B3LYP, MPW1PW91, and WB97XD functionals, as well as linear response polarisable continuum model; and estimation of global chemical reactivity descriptors, particularly the chemical hardness factor. The charges on pyrrolic and pyridinic nitrogen (the latter one in the quinolone ring of compound 8, as well as condensed Fukui functions) reveal a significant role of these atoms in potential interactions of azole ligand–protein binding pocket. The lowest negative value of free energy of solvation can be attributed to carbazole 6, whereas pyrazole 7 has the least negative value of this energy. Moreover, the HOMO–LUMO gap and chemical hardness show that carbazole 6 and indole 5 exist as soft molecules, while fused pyrazole 7 has hard character.

scholarly journals Calculation of the Global and Local Conceptual DFT Indices for the Prediction of the Chemical Reactivity Properties of Papuamides A–F Marine Drugs

Molecules ◽  
2019 ◽  
Vol 24 (18) ◽  
pp. 3312 ◽  
Author(s):  
Norma Flores-Holguín ◽  
Juan Frau ◽  
Daniel Glossman-Mitnik
Keyword(s):  
Density Functional Theory ◽  
Active Sites ◽  
Density Functional ◽  
Chemical Reactivity ◽  
Active Regions ◽  
Conceptual Density Functional Theory ◽  
Functional Theory ◽  
Reactivity Descriptors ◽  
Global And Local

A well-behaved model chemistry previously validated for the study of the chemical reactivity of peptides was considered for the calculation of the molecular properties and structures of the Papuamide family of marine peptides. A methodology based on Conceptual Density Functional Theory (CDFT) was chosen for the determination of the reactivity descriptors. The molecular active sites were associated with the active regions of the molecules related to the nucleophilic and electrophilic Parr functions. Finally, the drug-likenesses and the bioactivity scores for the Papuamide peptides were predicted through a homology methodology relating them with the calculated reactivity descriptors, while other properties such as the pKas were determined following a methodology developed by our group.

scholarly journals Virtual Screening of Marine Natural Compounds by Means of Chemoinformatics and CDFT-Based Computational Peptidology

Marine Drugs ◽  
2020 ◽  
Vol 18 (9) ◽  
pp. 478
Author(s):  
Norma Flores-Holguín ◽  
Juan Frau ◽  
Daniel Glossman-Mitnik
Keyword(s):  
Virtual Screening ◽  
Density Functional ◽  
Computational Study ◽  
Molecular System ◽  
Chemical Reactivity ◽  
Conceptual Density Functional Theory ◽  
Qsar Studies ◽  
Biological Targets ◽  
Global And Local

This work presents the results of a computational study of the chemical reactivity and bioactivity properties of the members of the theopapuamides A-D family of marine peptides by making use of our proposed methodology named Computational Peptidology (CP) that has been successfully considered in previous studies of this kind of molecular system. CP allows for the determination of the global and local descriptors that come from Conceptual Density Functional Theory (CDFT) that can give an idea about the chemical reactivity properties of the marine natural products under study, which are expected to be related to their bioactivity. At the same time, the validity of the procedure based on the adoption of the KID (Koopmans In DFT) technique, as well as the MN12SX/Def2TZVP/H2O model chemistry is successfully verified. Together with several chemoinformatic tools that can be used to improve the process of virtual screening, some additional properties of these marine peptides are identified related to their ability to behave as useful drugs. With the further objective of analyzing their bioactivity, some useful parameters for future QSAR studies, their predicted biological targets, and the ADMET (Absorption, Distribution, Metabolism, Excretion and Toxicity) parameters related to the theopapuamides A-D pharmacokinetics are also reported.

New molecular target insights about protein kinases of the Plasmodium falciparum. Using molecular docking and DFT-based reactivity descriptors

2017 ◽  
Vol 16 (08) ◽  
pp. 1750076 ◽  
Author(s):  
Alejandro Morales-Bayuelo
Keyword(s):  
Plasmodium Falciparum ◽  
Protein Kinases ◽  
Density Functional ◽  
Chemical Potential ◽  
Chemical Reactivity ◽  
Binding Pocket ◽  
Fukui Functions ◽  
Reactivity Descriptors ◽  
Chemical Reactivity Descriptors ◽  
Hinge Zone

Currently, there is increasing interest in the potential of malaria inhibitors in Plasmodium falciparum activity. In this work, is propose a possible alternative to classifying 154 antimalarials, with P. falciparum activity. These antimalarials were synthesized by the Chibale’s group ( http://www.kellychibaleresearch.uct.ac.za/ ), with the goal of finding new insights on the binding pocket of the protein kinase PfPK5, PfPK7, PfCDPK1, PfCDPK4, PfMAP1, and PfPK6 of the malaria parasite. However, there is only information about crystallography of PfPK5 and PfPK7. The protein kinases PfCDPK1, PfCDPK4, PfMAP1, and PfPK6 were modeled using molecular homology. The validation used shows that our homology models can be an alternative for the protein kinases from P. falciparum, unknown today. The antimalarials were classified by taking into account the interactions in the hinge zone. These ligands bind to the kinase through the formation of one of two hydrogen bonds, with the backbone residues of the hinge region connecting the kinase N- and C-terminal loops. These interactions were supported by a reactivity chemistry analysis, using global chemical reactivity descriptors such as chemical potential, hardness, softness, electrophilicity, and the Fukui functions as local reactivity descriptors, within the Density Functional Theory (DFT) context.

Sign in / Sign up

Export Citation Format

Share Document