A self-conformation-aware pre-training framework for molecular property prediction with substructure interpretability
Abstract The major challenges in drug development stem from frequent structure-activity cliffs and unknown drug properties, which are expensive and time-consuming to estimate, contributing to a high rate of failures and substantial unavoidable costs in the clinical phases. Herein, we propose the sel...
Saved in:
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-05-01
|
| Series: | Nature Communications |
| Online Access: | https://doi.org/10.1038/s41467-025-59634-0 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850154618577747968 |
|---|---|
| author | Jianbo Qiao Junru Jin Ding Wang Saisai Teng Junyu Zhang Xuetong Yang Yuhang Liu Yu Wang Lizhen Cui Quan Zou Ran Su Leyi Wei |
| author_facet | Jianbo Qiao Junru Jin Ding Wang Saisai Teng Junyu Zhang Xuetong Yang Yuhang Liu Yu Wang Lizhen Cui Quan Zou Ran Su Leyi Wei |
| author_sort | Jianbo Qiao |
| collection | DOAJ |
| description | Abstract The major challenges in drug development stem from frequent structure-activity cliffs and unknown drug properties, which are expensive and time-consuming to estimate, contributing to a high rate of failures and substantial unavoidable costs in the clinical phases. Herein, we propose the self-conformation-aware graph transformer (SCAGE), an innovative deep learning architecture pretrained with approximately 5 million drug-like compounds for molecular property prediction. Notably, we develop a multitask pretraining framework, which incorporates four supervised and unsupervised tasks: molecular fingerprint prediction, functional group prediction using chemical prior information, 2D atomic distance prediction, and 3D bond angle prediction, covering aspects from molecular structures to functions. It enables learning comprehensive conformation-aware prior knowledge, thereby enhancing its generalization across various molecular property tasks. Moreover, we design a data-driven multiscale conformational learning strategy that effectively guides the model in understanding and representing atomic relationships at the molecular conformational scale. SCAGE achieves significant performance improvements across 9 molecular properties and 30 structure-activity cliff benchmarks. Case studies demonstrate that SCAGE accurately captures crucial functional groups at the atomic level, which are closely associated with molecular activity, providing valuable insights into quantitative structure-activity relationships. |
| format | Article |
| id | doaj-art-b46a2b4e268d41e89f54a978cd6ee4f3 |
| institution | OA Journals |
| issn | 2041-1723 |
| language | English |
| publishDate | 2025-05-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Nature Communications |
| spelling | doaj-art-b46a2b4e268d41e89f54a978cd6ee4f32025-08-20T02:25:16ZengNature PortfolioNature Communications2041-17232025-05-0116111610.1038/s41467-025-59634-0A self-conformation-aware pre-training framework for molecular property prediction with substructure interpretabilityJianbo Qiao0Junru Jin1Ding Wang2Saisai Teng3Junyu Zhang4Xuetong Yang5Yuhang Liu6Yu Wang7Lizhen Cui8Quan Zou9Ran Su10Leyi Wei11School of Software, Shandong UniversitySchool of Software, Shandong UniversitySchool of Software, Shandong UniversitySchool of Software, Shandong UniversitySchool of Software, Shandong UniversitySchool of Software, Shandong UniversityFaculty of Applied Sciences, Macao Polytechnic UniversitySchool of Software, Shandong UniversitySchool of Software, Shandong UniversityInstitute of Fundamental and Frontier Sciences, University of Electronic Science and Technology of ChinaCollege of Intelligence and Computing, Tianjin UniversityFaculty of Applied Sciences, Macao Polytechnic UniversityAbstract The major challenges in drug development stem from frequent structure-activity cliffs and unknown drug properties, which are expensive and time-consuming to estimate, contributing to a high rate of failures and substantial unavoidable costs in the clinical phases. Herein, we propose the self-conformation-aware graph transformer (SCAGE), an innovative deep learning architecture pretrained with approximately 5 million drug-like compounds for molecular property prediction. Notably, we develop a multitask pretraining framework, which incorporates four supervised and unsupervised tasks: molecular fingerprint prediction, functional group prediction using chemical prior information, 2D atomic distance prediction, and 3D bond angle prediction, covering aspects from molecular structures to functions. It enables learning comprehensive conformation-aware prior knowledge, thereby enhancing its generalization across various molecular property tasks. Moreover, we design a data-driven multiscale conformational learning strategy that effectively guides the model in understanding and representing atomic relationships at the molecular conformational scale. SCAGE achieves significant performance improvements across 9 molecular properties and 30 structure-activity cliff benchmarks. Case studies demonstrate that SCAGE accurately captures crucial functional groups at the atomic level, which are closely associated with molecular activity, providing valuable insights into quantitative structure-activity relationships.https://doi.org/10.1038/s41467-025-59634-0 |
| spellingShingle | Jianbo Qiao Junru Jin Ding Wang Saisai Teng Junyu Zhang Xuetong Yang Yuhang Liu Yu Wang Lizhen Cui Quan Zou Ran Su Leyi Wei A self-conformation-aware pre-training framework for molecular property prediction with substructure interpretability Nature Communications |
| title | A self-conformation-aware pre-training framework for molecular property prediction with substructure interpretability |
| title_full | A self-conformation-aware pre-training framework for molecular property prediction with substructure interpretability |
| title_fullStr | A self-conformation-aware pre-training framework for molecular property prediction with substructure interpretability |
| title_full_unstemmed | A self-conformation-aware pre-training framework for molecular property prediction with substructure interpretability |
| title_short | A self-conformation-aware pre-training framework for molecular property prediction with substructure interpretability |
| title_sort | self conformation aware pre training framework for molecular property prediction with substructure interpretability |
| url | https://doi.org/10.1038/s41467-025-59634-0 |
| work_keys_str_mv | AT jianboqiao aselfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT junrujin aselfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT dingwang aselfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT saisaiteng aselfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT junyuzhang aselfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT xuetongyang aselfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT yuhangliu aselfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT yuwang aselfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT lizhencui aselfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT quanzou aselfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT ransu aselfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT leyiwei aselfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT jianboqiao selfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT junrujin selfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT dingwang selfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT saisaiteng selfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT junyuzhang selfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT xuetongyang selfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT yuhangliu selfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT yuwang selfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT lizhencui selfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT quanzou selfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT ransu selfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability AT leyiwei selfconformationawarepretrainingframeworkformolecularpropertypredictionwithsubstructureinterpretability |