Q-DFTNet: A chemistry-informed neural network framework for predicting molecular dipole moments via DFT-Driven QM9 data
This study presents Q-DFTNet, a chemistry-informed neural network (ChINN) framework designed to benchmark graph neural networks (GNNs) for dipole moment prediction using the QM9 dataset. Seven GNN architectures, GCN, GIN, GraphConv, GATConv, GATNet, SAGEConv, and GIN+EdgeConv, were trained for 100 e...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | en |
| Published: |
John Wiley and Sons Inc.
2025
|
| Subjects: | |
| Online Access: | https://umpir.ump.edu.my/id/eprint/47149/1/Q-DFTNet_A%20chemistry-informed%20neural%20network%20framework.pdf https://doi.org/10.1002/jcc.70206 https://umpir.ump.edu.my/id/eprint/47149/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1858062138545799168 |
|---|---|
| author | Wayo, Dennis Delali Kwesi Mohd Zulkifli, Mohamad Noor Ganji, Masoud Darvish Saporetti, Camila M. Goliatt, Leonardo |
| author_facet | Wayo, Dennis Delali Kwesi Mohd Zulkifli, Mohamad Noor Ganji, Masoud Darvish Saporetti, Camila M. Goliatt, Leonardo |
| author_sort | Wayo, Dennis Delali Kwesi |
| building | UMPSA Library |
| collection | Institutional Repository |
| content_provider | Universiti Malaysia Pahang Al-Sultan Abdullah |
| content_source | UMPSA Institutional Repository |
| continent | Asia |
| country | Malaysia |
| description | This study presents Q-DFTNet, a chemistry-informed neural network (ChINN) framework designed to benchmark graph neural networks (GNNs) for dipole moment prediction using the QM9 dataset. Seven GNN architectures, GCN, GIN, GraphConv, GATConv, GATNet, SAGEConv, and GIN+EdgeConv, were trained for 100 epochs and evaluated across performance and interpretability metrics. GraphConv achieved the lowest test MSE (0.7054), MAE (0.6196), and the highest R2 (0.6513) with only 16.5k trainable parameters, confirming its optimal accuracy-complexity trade-off. GIN+EdgeConv followed closely with MSE of 0.7386, MAE of 0.6332, and R2 of 0.6349, leveraging edge-awareness for enhanced expressivity. In contrast, attention-based models like GATConv and GATNet underperformed, with test MSEs of 0.9667 and 1.0096, and R2 values of 0.5221 and 0.5009, despite their higher complexity (43.5k and 37.3k parameters). Latent space analysis via t-SNE, PCA, and UMAP showed superior cluster separability for GraphConv, GIN+EdgeConv, and GCN. Clustering metrics corroborated these observations: GraphConv achieved a Silhouette Score of 0.4665, a Davies–Bouldin Index of 0.7111, and a Calinski–Harabasz Score of 1278.40. Cluster-wise molecular dipole means for GIN+EdgeConv ranged from 2.6221 to 2.9606 Debye, reflecting high semantic coherence. Residual analysis and QQ plots confirmed that models with lower MSEs also had near-Gaussian error distributions, enhancing interpretability. |
| format | Article |
| id | my.ump.umpir-47149 |
| institution | Universiti Malaysia Pahang |
| language | en |
| publishDate | 2025 |
| publisher | John Wiley and Sons Inc. |
| record_format | eprints |
| spelling | my.ump.umpir-471492026-02-11T06:42:41Z https://umpir.ump.edu.my/id/eprint/47149/ Q-DFTNet: A chemistry-informed neural network framework for predicting molecular dipole moments via DFT-Driven QM9 data Wayo, Dennis Delali Kwesi Mohd Zulkifli, Mohamad Noor Ganji, Masoud Darvish Saporetti, Camila M. Goliatt, Leonardo TP Chemical technology This study presents Q-DFTNet, a chemistry-informed neural network (ChINN) framework designed to benchmark graph neural networks (GNNs) for dipole moment prediction using the QM9 dataset. Seven GNN architectures, GCN, GIN, GraphConv, GATConv, GATNet, SAGEConv, and GIN+EdgeConv, were trained for 100 epochs and evaluated across performance and interpretability metrics. GraphConv achieved the lowest test MSE (0.7054), MAE (0.6196), and the highest R2 (0.6513) with only 16.5k trainable parameters, confirming its optimal accuracy-complexity trade-off. GIN+EdgeConv followed closely with MSE of 0.7386, MAE of 0.6332, and R2 of 0.6349, leveraging edge-awareness for enhanced expressivity. In contrast, attention-based models like GATConv and GATNet underperformed, with test MSEs of 0.9667 and 1.0096, and R2 values of 0.5221 and 0.5009, despite their higher complexity (43.5k and 37.3k parameters). Latent space analysis via t-SNE, PCA, and UMAP showed superior cluster separability for GraphConv, GIN+EdgeConv, and GCN. Clustering metrics corroborated these observations: GraphConv achieved a Silhouette Score of 0.4665, a Davies–Bouldin Index of 0.7111, and a Calinski–Harabasz Score of 1278.40. Cluster-wise molecular dipole means for GIN+EdgeConv ranged from 2.6221 to 2.9606 Debye, reflecting high semantic coherence. Residual analysis and QQ plots confirmed that models with lower MSEs also had near-Gaussian error distributions, enhancing interpretability. John Wiley and Sons Inc. 2025-08-13 Article PeerReviewed pdf en https://umpir.ump.edu.my/id/eprint/47149/1/Q-DFTNet_A%20chemistry-informed%20neural%20network%20framework.pdf Wayo, Dennis Delali Kwesi and Mohd Zulkifli, Mohamad Noor and Ganji, Masoud Darvish and Saporetti, Camila M. and Goliatt, Leonardo (2025) Q-DFTNet: A chemistry-informed neural network framework for predicting molecular dipole moments via DFT-Driven QM9 data. Journal of Computational Chemistry, 46 (22). pp. 1-19. ISSN 0192-8651. (Published) https://doi.org/10.1002/jcc.70206 https://doi.org/10.1002/jcc.70206 https://doi.org/10.1002/jcc.70206 |
| spellingShingle | TP Chemical technology Wayo, Dennis Delali Kwesi Mohd Zulkifli, Mohamad Noor Ganji, Masoud Darvish Saporetti, Camila M. Goliatt, Leonardo Q-DFTNet: A chemistry-informed neural network framework for predicting molecular dipole moments via DFT-Driven QM9 data |
| title | Q-DFTNet: A chemistry-informed neural network framework for predicting molecular dipole moments via DFT-Driven QM9 data |
| title_full | Q-DFTNet: A chemistry-informed neural network framework for predicting molecular dipole moments via DFT-Driven QM9 data |
| title_fullStr | Q-DFTNet: A chemistry-informed neural network framework for predicting molecular dipole moments via DFT-Driven QM9 data |
| title_full_unstemmed | Q-DFTNet: A chemistry-informed neural network framework for predicting molecular dipole moments via DFT-Driven QM9 data |
| title_short | Q-DFTNet: A chemistry-informed neural network framework for predicting molecular dipole moments via DFT-Driven QM9 data |
| title_sort | q-dftnet: a chemistry-informed neural network framework for predicting molecular dipole moments via dft-driven qm9 data |
| topic | TP Chemical technology |
| url | https://umpir.ump.edu.my/id/eprint/47149/1/Q-DFTNet_A%20chemistry-informed%20neural%20network%20framework.pdf https://doi.org/10.1002/jcc.70206 https://umpir.ump.edu.my/id/eprint/47149/ https://doi.org/10.1002/jcc.70206 https://doi.org/10.1002/jcc.70206 |
| url_provider | http://umpir.ump.edu.my/ |