A Fast, Low-Cost and Simple Method for Predicting Atomic/Inter-Atomic Properties by Combining a Low Dimensional Deep Learning Model with a Fragment Based Graph Convolutional Network
Calculations with high accuracy for atomic and inter-atomic properties, such as nuclear magnetic resonance (NMR) spectroscopy and bond dissociation energies (BDEs) are valuable for pharmaceutical molecule structural analysis, drug exploration, and screening. It is important that these calculations s...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-12-01
|
| Series: | Crystals |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2073-4352/12/12/1740 |
| _version_ | 1797460752539320320 |
|---|---|
| author | Peng Gao Zonghang Liu Jie Zhang Jia-Ao Wang Graeme Henkelman |
| author_facet | Peng Gao Zonghang Liu Jie Zhang Jia-Ao Wang Graeme Henkelman |
| author_sort | Peng Gao |
| collection | DOAJ |
| description | Calculations with high accuracy for atomic and inter-atomic properties, such as nuclear magnetic resonance (NMR) spectroscopy and bond dissociation energies (BDEs) are valuable for pharmaceutical molecule structural analysis, drug exploration, and screening. It is important that these calculations should include relativistic effects, which are computationally expensive to treat. Non-relativistic calculations are less expensive but their results are less accurate. In this study, we present a computational framework for predicting atomic and inter-atomic properties by using machine-learning in a non-relativistic but accurate and computationally inexpensive framework. The accurate atomic and inter-atomic properties are obtained with a low dimensional deep neural network (DNN) embedded in a fragment-based graph convolutional neural network (F-GCN). The F-GCN acts as an atomic fingerprint generator that converts the atomistic local environments into data for the DNN, which improves the learning ability, resulting in accurate results as compared to experiments. Using this framework, the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mn>13</mn></msup></semantics></math></inline-formula>C/<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mn>1</mn></msup></semantics></math></inline-formula>H NMR chemical shifts of Nevirapine and phenol O–H BDEs are predicted to be in good agreement with experimental measurement. |
| first_indexed | 2024-03-09T17:10:29Z |
| format | Article |
| id | doaj.art-99184b905f9b45ba9a5656956e22de16 |
| institution | Directory Open Access Journal |
| issn | 2073-4352 |
| language | English |
| last_indexed | 2024-03-09T17:10:29Z |
| publishDate | 2022-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Crystals |
| spelling | doaj.art-99184b905f9b45ba9a5656956e22de162023-11-24T14:10:12ZengMDPI AGCrystals2073-43522022-12-011212174010.3390/cryst12121740A Fast, Low-Cost and Simple Method for Predicting Atomic/Inter-Atomic Properties by Combining a Low Dimensional Deep Learning Model with a Fragment Based Graph Convolutional NetworkPeng Gao0Zonghang Liu1Jie Zhang2Jia-Ao Wang3Graeme Henkelman4School of Chemistry and Molecular Bioscience, University of Wollongong, Wollongong, NSW 2500, AustraliaSchool of Chemistry and Chemical Engineering, University of Jinan, Jinan 250022, ChinaSchool of Chemical Engineering, East China University of Science and Technology, Shanghai 200237, ChinaDepartment of Chemistry, Oden Institute for Computational Engineering and Sciences, University of Texas at Austin, Austin, TX 78712, USADepartment of Chemistry, Oden Institute for Computational Engineering and Sciences, University of Texas at Austin, Austin, TX 78712, USACalculations with high accuracy for atomic and inter-atomic properties, such as nuclear magnetic resonance (NMR) spectroscopy and bond dissociation energies (BDEs) are valuable for pharmaceutical molecule structural analysis, drug exploration, and screening. It is important that these calculations should include relativistic effects, which are computationally expensive to treat. Non-relativistic calculations are less expensive but their results are less accurate. In this study, we present a computational framework for predicting atomic and inter-atomic properties by using machine-learning in a non-relativistic but accurate and computationally inexpensive framework. The accurate atomic and inter-atomic properties are obtained with a low dimensional deep neural network (DNN) embedded in a fragment-based graph convolutional neural network (F-GCN). The F-GCN acts as an atomic fingerprint generator that converts the atomistic local environments into data for the DNN, which improves the learning ability, resulting in accurate results as compared to experiments. Using this framework, the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mn>13</mn></msup></semantics></math></inline-formula>C/<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mn>1</mn></msup></semantics></math></inline-formula>H NMR chemical shifts of Nevirapine and phenol O–H BDEs are predicted to be in good agreement with experimental measurement.https://www.mdpi.com/2073-4352/12/12/1740quantum mechanicsneural networkNMRbond dissociation energymachine-learning |
| spellingShingle | Peng Gao Zonghang Liu Jie Zhang Jia-Ao Wang Graeme Henkelman A Fast, Low-Cost and Simple Method for Predicting Atomic/Inter-Atomic Properties by Combining a Low Dimensional Deep Learning Model with a Fragment Based Graph Convolutional Network Crystals quantum mechanics neural network NMR bond dissociation energy machine-learning |
| title | A Fast, Low-Cost and Simple Method for Predicting Atomic/Inter-Atomic Properties by Combining a Low Dimensional Deep Learning Model with a Fragment Based Graph Convolutional Network |
| title_full | A Fast, Low-Cost and Simple Method for Predicting Atomic/Inter-Atomic Properties by Combining a Low Dimensional Deep Learning Model with a Fragment Based Graph Convolutional Network |
| title_fullStr | A Fast, Low-Cost and Simple Method for Predicting Atomic/Inter-Atomic Properties by Combining a Low Dimensional Deep Learning Model with a Fragment Based Graph Convolutional Network |
| title_full_unstemmed | A Fast, Low-Cost and Simple Method for Predicting Atomic/Inter-Atomic Properties by Combining a Low Dimensional Deep Learning Model with a Fragment Based Graph Convolutional Network |
| title_short | A Fast, Low-Cost and Simple Method for Predicting Atomic/Inter-Atomic Properties by Combining a Low Dimensional Deep Learning Model with a Fragment Based Graph Convolutional Network |
| title_sort | fast low cost and simple method for predicting atomic inter atomic properties by combining a low dimensional deep learning model with a fragment based graph convolutional network |
| topic | quantum mechanics neural network NMR bond dissociation energy machine-learning |
| url | https://www.mdpi.com/2073-4352/12/12/1740 |
| work_keys_str_mv | AT penggao afastlowcostandsimplemethodforpredictingatomicinteratomicpropertiesbycombiningalowdimensionaldeeplearningmodelwithafragmentbasedgraphconvolutionalnetwork AT zonghangliu afastlowcostandsimplemethodforpredictingatomicinteratomicpropertiesbycombiningalowdimensionaldeeplearningmodelwithafragmentbasedgraphconvolutionalnetwork AT jiezhang afastlowcostandsimplemethodforpredictingatomicinteratomicpropertiesbycombiningalowdimensionaldeeplearningmodelwithafragmentbasedgraphconvolutionalnetwork AT jiaaowang afastlowcostandsimplemethodforpredictingatomicinteratomicpropertiesbycombiningalowdimensionaldeeplearningmodelwithafragmentbasedgraphconvolutionalnetwork AT graemehenkelman afastlowcostandsimplemethodforpredictingatomicinteratomicpropertiesbycombiningalowdimensionaldeeplearningmodelwithafragmentbasedgraphconvolutionalnetwork AT penggao fastlowcostandsimplemethodforpredictingatomicinteratomicpropertiesbycombiningalowdimensionaldeeplearningmodelwithafragmentbasedgraphconvolutionalnetwork AT zonghangliu fastlowcostandsimplemethodforpredictingatomicinteratomicpropertiesbycombiningalowdimensionaldeeplearningmodelwithafragmentbasedgraphconvolutionalnetwork AT jiezhang fastlowcostandsimplemethodforpredictingatomicinteratomicpropertiesbycombiningalowdimensionaldeeplearningmodelwithafragmentbasedgraphconvolutionalnetwork AT jiaaowang fastlowcostandsimplemethodforpredictingatomicinteratomicpropertiesbycombiningalowdimensionaldeeplearningmodelwithafragmentbasedgraphconvolutionalnetwork AT graemehenkelman fastlowcostandsimplemethodforpredictingatomicinteratomicpropertiesbycombiningalowdimensionaldeeplearningmodelwithafragmentbasedgraphconvolutionalnetwork |