NeuTox 2.0: A hybrid deep learning architecture for screening potential neurotoxicity of chemicals based on multimodal feature fusion
Chemically induced neurotoxicity is a critical aspect of chemical safety assessment. Traditional and costly experimental methods call for the development of high-throughput virtual screening. However, the small datasets of neurotoxicity have limited the application of advanced deep learning techniqu...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-01-01
|
| Series: | Environment International |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S0160412024008316 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832589990935134208 |
|---|---|
| author | Xudi Pang Xuejun He Ying Yang Ling Wang Yuzhen Sun Huiming Cao Yong Liang |
| author_facet | Xudi Pang Xuejun He Ying Yang Ling Wang Yuzhen Sun Huiming Cao Yong Liang |
| author_sort | Xudi Pang |
| collection | DOAJ |
| description | Chemically induced neurotoxicity is a critical aspect of chemical safety assessment. Traditional and costly experimental methods call for the development of high-throughput virtual screening. However, the small datasets of neurotoxicity have limited the application of advanced deep learning techniques. The current study developed a hybrid deep learning architecture, NeuTox 2.0, through multimodal feature fusion for enhanced prediction accuracy and generalization ability. We incorporated transfer learning based on self-supervised learning, graph neural networks, and molecular fingerprints/descriptors. Four datasets were used to profile neurotoxicity; these were related to blood–brain barrier permeability, neuronal cytotoxicity, microelectrode array–based neural activity, and mammalian neurotoxicity. Comprehensive performance evaluations demonstrated that NeuTox 2.0 has relatively higher predictive capability across all statistical metrics. Specifically, NeuTox 2.0 exhibits remarkable performance in three of the four datasets. In the BBB dataset, although it does not outperform the PaDEL descriptor model, its performance closely approximates that of the top single-modal model. The ablation experiments indicated that NeuTox 2.0 can learn the deeper structural differences of molecules from various feature extractions and capture complex interactions and mapping relationships between various modalities, thereby improving performance for neurotoxicity prediction. Evaluations of anti-noise ability indicated that NeuTox 2.0 has excellent noise resistance relative to traditional machine learning. We applied the NeuTox 2.0 model to predict the neurotoxicity of 315,790 compounds in the REACH database. The results showed that 701 compounds exhibited potential neurotoxicity in the four neurotoxicity-related predictions. In conclusion, NeuTox 2.0 can be used as an efficient tool for early neurotoxicity screening of environmental chemicals. |
| format | Article |
| id | doaj-art-91221bbcdbf941fc80b598548795d592 |
| institution | Kabale University |
| issn | 0160-4120 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Environment International |
| spelling | doaj-art-91221bbcdbf941fc80b598548795d5922025-01-24T04:44:12ZengElsevierEnvironment International0160-41202025-01-01195109244NeuTox 2.0: A hybrid deep learning architecture for screening potential neurotoxicity of chemicals based on multimodal feature fusionXudi Pang0Xuejun He1Ying Yang2Ling Wang3Yuzhen Sun4Huiming Cao5Yong Liang6Hubei Key Laboratory of Environmental and Health Effects of Persistent Toxic Substances, School of Environment and Health, Jianghan University, Wuhan 430056, ChinaHubei Key Laboratory of Environmental and Health Effects of Persistent Toxic Substances, School of Environment and Health, Jianghan University, Wuhan 430056, ChinaHubei Key Laboratory of Environmental and Health Effects of Persistent Toxic Substances, School of Environment and Health, Jianghan University, Wuhan 430056, ChinaHubei Key Laboratory of Environmental and Health Effects of Persistent Toxic Substances, School of Environment and Health, Jianghan University, Wuhan 430056, ChinaHubei Key Laboratory of Environmental and Health Effects of Persistent Toxic Substances, School of Environment and Health, Jianghan University, Wuhan 430056, ChinaCorresponding author.; Hubei Key Laboratory of Environmental and Health Effects of Persistent Toxic Substances, School of Environment and Health, Jianghan University, Wuhan 430056, ChinaHubei Key Laboratory of Environmental and Health Effects of Persistent Toxic Substances, School of Environment and Health, Jianghan University, Wuhan 430056, ChinaChemically induced neurotoxicity is a critical aspect of chemical safety assessment. Traditional and costly experimental methods call for the development of high-throughput virtual screening. However, the small datasets of neurotoxicity have limited the application of advanced deep learning techniques. The current study developed a hybrid deep learning architecture, NeuTox 2.0, through multimodal feature fusion for enhanced prediction accuracy and generalization ability. We incorporated transfer learning based on self-supervised learning, graph neural networks, and molecular fingerprints/descriptors. Four datasets were used to profile neurotoxicity; these were related to blood–brain barrier permeability, neuronal cytotoxicity, microelectrode array–based neural activity, and mammalian neurotoxicity. Comprehensive performance evaluations demonstrated that NeuTox 2.0 has relatively higher predictive capability across all statistical metrics. Specifically, NeuTox 2.0 exhibits remarkable performance in three of the four datasets. In the BBB dataset, although it does not outperform the PaDEL descriptor model, its performance closely approximates that of the top single-modal model. The ablation experiments indicated that NeuTox 2.0 can learn the deeper structural differences of molecules from various feature extractions and capture complex interactions and mapping relationships between various modalities, thereby improving performance for neurotoxicity prediction. Evaluations of anti-noise ability indicated that NeuTox 2.0 has excellent noise resistance relative to traditional machine learning. We applied the NeuTox 2.0 model to predict the neurotoxicity of 315,790 compounds in the REACH database. The results showed that 701 compounds exhibited potential neurotoxicity in the four neurotoxicity-related predictions. In conclusion, NeuTox 2.0 can be used as an efficient tool for early neurotoxicity screening of environmental chemicals.http://www.sciencedirect.com/science/article/pii/S0160412024008316Deep learningMultimodalTransfer learningNeurotoxicity |
| spellingShingle | Xudi Pang Xuejun He Ying Yang Ling Wang Yuzhen Sun Huiming Cao Yong Liang NeuTox 2.0: A hybrid deep learning architecture for screening potential neurotoxicity of chemicals based on multimodal feature fusion Environment International Deep learning Multimodal Transfer learning Neurotoxicity |
| title | NeuTox 2.0: A hybrid deep learning architecture for screening potential neurotoxicity of chemicals based on multimodal feature fusion |
| title_full | NeuTox 2.0: A hybrid deep learning architecture for screening potential neurotoxicity of chemicals based on multimodal feature fusion |
| title_fullStr | NeuTox 2.0: A hybrid deep learning architecture for screening potential neurotoxicity of chemicals based on multimodal feature fusion |
| title_full_unstemmed | NeuTox 2.0: A hybrid deep learning architecture for screening potential neurotoxicity of chemicals based on multimodal feature fusion |
| title_short | NeuTox 2.0: A hybrid deep learning architecture for screening potential neurotoxicity of chemicals based on multimodal feature fusion |
| title_sort | neutox 2 0 a hybrid deep learning architecture for screening potential neurotoxicity of chemicals based on multimodal feature fusion |
| topic | Deep learning Multimodal Transfer learning Neurotoxicity |
| url | http://www.sciencedirect.com/science/article/pii/S0160412024008316 |
| work_keys_str_mv | AT xudipang neutox20ahybriddeeplearningarchitectureforscreeningpotentialneurotoxicityofchemicalsbasedonmultimodalfeaturefusion AT xuejunhe neutox20ahybriddeeplearningarchitectureforscreeningpotentialneurotoxicityofchemicalsbasedonmultimodalfeaturefusion AT yingyang neutox20ahybriddeeplearningarchitectureforscreeningpotentialneurotoxicityofchemicalsbasedonmultimodalfeaturefusion AT lingwang neutox20ahybriddeeplearningarchitectureforscreeningpotentialneurotoxicityofchemicalsbasedonmultimodalfeaturefusion AT yuzhensun neutox20ahybriddeeplearningarchitectureforscreeningpotentialneurotoxicityofchemicalsbasedonmultimodalfeaturefusion AT huimingcao neutox20ahybriddeeplearningarchitectureforscreeningpotentialneurotoxicityofchemicalsbasedonmultimodalfeaturefusion AT yongliang neutox20ahybriddeeplearningarchitectureforscreeningpotentialneurotoxicityofchemicalsbasedonmultimodalfeaturefusion |