pith. sign in

arxiv: 2605.16799 · v1 · pith:JMG6TEZMnew · submitted 2026-05-16 · 💻 cs.LG · cs.AI

Cross-Domain Molecular Relational Learning: Leveraging Chemical Structure-Activity Analysis

Peiliang Zhang , Jingling Yuan , Shiqing Wu , Mengqing Hu , Chao Che , Yongjun Zhu , Lin Li This is my paper

Pith reviewed 2026-05-19 21:19 UTC · model grok-4.3

classification 💻 cs.LG cs.AI
keywords cross-domain learningmolecular relational learningdomain adversarial traininggradient reversalstructure-activity analysistransfer learningmolecular representation
0
0 comments X

The pith

DisTrans adapts molecular representations across domains by reversing gradients on substructure topological differences to keep them transferable.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper seeks to overcome the limitation of existing molecular relational learning methods that only work within a single domain by developing a cross-domain approach. It introduces DisTrans, which uses structure-activity analysis to handle transfers between different molecular datasets, such as structures and images. A sympathetic reader would care if this enables more flexible use of molecular models in real-world scenarios where data distributions vary, like in drug discovery across different chemical libraries. The method combines adversarial training with semantic alignment to maintain performance despite domain shifts.

Core claim

The central discovery is that a Domain Adversarial Training Network with Structural-Semantic Transfer Discrepancy (DisTrans) can optimize cross-domain adaptive representations for molecular structures and visual images. By employing gradient reversal based on substructure topological discrepancies, the model learns domain dependence and generates domain-separable structural representations that adapt to target domain patterns. Additionally, a cross-domain representation guidance mechanism aligns functional-group semantic information to learn consistency across domains. Experiments in two cross-domain strategies show it outperforms 16 baselines even with high inter-domain discrepancy.

What carries the argument

Gradient reversal strategy applied to substructure topological discrepancies between domains, which learns domain dependence and produces domain-separable yet transferable molecular structural representations.

If this is right

  • Cross-domain molecular relational learning becomes feasible without discarding task-relevant features.
  • The model adapts to structural adjacency patterns in the target domain.
  • Functional-group semantic information is aligned between source and target domains.
  • Performance remains satisfactory even under pronounced inter-domain discrepancy.
  • Superior results compared to 16 baseline methods in typical cross-domain setups.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • This could allow combining molecular graph data with image-based representations in a more robust way for multi-modal learning.
  • Similar gradient reversal techniques might help in other domains where topological structures shift, such as in protein interaction networks.
  • Testing the method on real-world molecular datasets with natural domain shifts could validate broader applicability in cheminformatics.

Load-bearing premise

Substructure topological discrepancies between source and target domains can be leveraged via gradient reversal to produce domain-separable yet transferable structural representations without discarding task-relevant molecular features.

What would settle it

If visualizations or metrics show that the structural representations lose either domain separability or predictive power for molecular activity when gradient reversal is applied, or if performance degrades sharply as inter-domain discrepancy increases beyond tested levels.

Figures

Figures reproduced from arXiv: 2605.16799 by Chao Che, Jingling Yuan, Lin Li, Mengqing Hu, Peiliang Zhang, Shiqing Wu, Yongjun Zhu.

Figure 1
Figure 1. Figure 1: The comparison of structural differences between [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The framework of DisTrans. (a) TRegCross serves as the feature extractor to capture molecular representation. (b) [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The performance and representation differences of different transfer strategies. (a) and (b) present the ACC in IDHT. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The cross-domain adaptability prediction perfor [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The predictive Performance of OnlyTop, OnlySem, [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The visualization results of cross-domain representational discrepancies among heterogeneous molecular types. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: The visualization results of molecular representa [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
read the original abstract

Recent advances in molecular representation integrates molecular topological and visual modalities, opening new avenues for precise Molecular Relational Learning (MRL). Existing MRL methods focus on intra-domain modeling, and their inherent domain-closed effect limits applicability to molecular science, particularly in elucidating cross-domain interaction mechanisms. Consequently, the imperative for Cross-Domain Molecular Relational Learning has become increasingly pressing. Benefiting from structure-activity analysis, we propose the Domain Adversarial Training Network with Structural-Semantic Transfer Discrepancy (DisTrans) to optimize cross-domain adaptive representation for molecular structures and visual images. 1) We employ the gradient reversal strategy based on substructure topological discrepancies between domains to learn the domain dependence of molecular structures. This strategy guides the model to adapt to the structural adjacency patterns in the target domain, generating domain-separable structural representations. 2) We apply the cross-domain representation guidance mechanism to align the functional-group semantic information between the source and target domains, learning cross-domain consistency information. The experimental results in two typical cross-domain strategies demonstrate that DisTrans outperforms 16 baseline methods, maintaining satisfactory performance even under pronounced inter-domain discrepancy.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript proposes DisTrans, a Domain Adversarial Training Network with Structural-Semantic Transfer Discrepancy, for cross-domain molecular relational learning. It employs gradient reversal on substructure topological discrepancies between source and target domains to learn domain dependence and produce domain-separable structural representations that adapt to target adjacency patterns, combined with a cross-domain representation guidance mechanism to align functional-group semantic information. Experiments in two typical cross-domain strategies are reported to show outperformance over 16 baselines while maintaining performance under large inter-domain discrepancy.

Significance. If the central claims hold after clarification, the work would provide a useful empirical framework for handling domain shifts in molecular representation learning, particularly by combining adversarial adaptation with semantic alignment for structure-activity tasks. This could support more robust transfer in scenarios with limited labeled target-domain molecular data.

major comments (1)
  1. [Abstract / Proposed Method] Abstract and method description: The statement that gradient reversal based on substructure topological discrepancies 'guides the model to adapt to the structural adjacency patterns in the target domain, generating domain-separable structural representations' conflicts with the standard formulation of gradient reversal (as in DANN), which minimizes domain discrepancy to yield invariant rather than separable features. The manuscript must specify the precise reversal direction, loss formulation, and how separability is achieved without discarding task-relevant features; this mechanism is load-bearing for the adaptation claim.
minor comments (2)
  1. [Abstract] Abstract: The two 'typical cross-domain strategies' are referenced but not named or described; early clarification would improve readability.
  2. [Abstract] The abstract asserts outperformance over 16 baselines without any numerical results, dataset sizes, or statistical details; the full manuscript should ensure these appear with error bars and ablations in the experimental section.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. The major comment raises a valid point about potential ambiguity in our description of the gradient reversal mechanism. We address it directly below and will revise the manuscript accordingly to improve clarity without altering the underlying technical approach.

read point-by-point responses

  1. Referee: Abstract and method description: The statement that gradient reversal based on substructure topological discrepancies 'guides the model to adapt to the structural adjacency patterns in the target domain, generating domain-separable structural representations' conflicts with the standard formulation of gradient reversal (as in DANN), which minimizes domain discrepancy to yield invariant rather than separable features. The manuscript must specify the precise reversal direction, loss formulation, and how separability is achieved without discarding task-relevant features; this mechanism is load-bearing for the adaptation claim.

    Authors: We agree that the phrasing 'domain-separable structural representations' is imprecise and risks misinterpretation relative to standard domain-adversarial training. In DisTrans, gradient reversal is applied to the domain classifier loss (with negative gradient flow) to encourage the structural encoder to minimize domain discrepancy arising from substructure topological differences, thereby producing domain-invariant features. These invariant representations are then combined with the cross-domain representation guidance mechanism to align functional-group semantics and adapt to target-domain adjacency patterns for the downstream relational learning task. Task-relevant features are preserved by the joint optimization with the primary molecular relational objective. We will revise the abstract and Section 3 to include the exact loss formulation (including the reversal coefficient and domain classification objective), clarify that the representations are domain-invariant rather than separable, and add a brief derivation showing how this avoids discarding task-relevant information. This revision will be accompanied by an updated Figure 2 for visual clarity. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical method with independent experimental validation

full rationale

The paper proposes an empirical architecture (DisTrans) combining gradient reversal on substructure discrepancies with semantic alignment, then reports outperformance on cross-domain tasks against 16 baselines. No equations, derivations, or first-principles predictions appear in the provided text that reduce claimed results to quantities defined by the method's own fitted parameters or self-citations. Performance claims rest on external benchmark comparisons rather than internal self-reference, satisfying the self-contained criterion.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The approach rests on standard domain-adaptation assumptions plus the untested premise that topological and semantic signals can be disentangled for transfer; no free parameters or invented physical entities are explicitly introduced in the abstract.

axioms (1)
  • domain assumption Molecular structures admit consistent topological and visual representations whose discrepancies can be exploited for domain adaptation.
    Invoked when the gradient-reversal strategy is applied to substructure patterns.
invented entities (1)
  • DisTrans network no independent evidence
    purpose: Cross-domain adaptive representation for molecular structures and images
    New model component introduced to solve the stated problem; no external falsifiable prediction supplied.

pith-pipeline@v0.9.0 · 5740 in / 1224 out tokens · 36442 ms · 2026-05-19T21:19:33.371383+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

47 extracted references · 47 canonical work pages

  1. [1]

    Yu, and Edwin R

    Lu Bai, Lixin Cui, Yue Wang, Ming Li, Jing Li, Philip S. Yu, and Edwin R. Hancock

    work page

  2. [2]

    HAQJSK: Hierarchical-Aligned Quantum Jensen-Shannon Kernels for Graph Classification.IEEE Transactions on Knowledge and Data Engineering (2024), 1–14

    work page 2024

  3. [3]

    Mohamed Ishmael Belghazi, Aristide Baratin, Sai Rajeshwar, Sherjil Ozair, Yoshua Bengio, Aaron Courville, and Devon Hjelm. 2018. Mutual Information Neural Es- timation. InProceedings of the 35th International Conference on Machine Learning. 531–540

    work page 2018

  4. [4]

    2015.In Silico Medicinal Chemistry: Computational Methods to Support Drug Design

    Nathan Brown. 2015.In Silico Medicinal Chemistry: Computational Methods to Support Drug Design. Royal Society of Chemistry

    work page 2015

  5. [5]

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xi- aohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. 2020. An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale. InProceedings of the 8th International Conference on Learning Representations. Cross-Domain ...

    work page 2020

  6. [6]

    Wenjie Du, Shuai Zhang, Jun Xia Di Wu, Ziyuan Zhao, Junfeng Fang, and Yang Wang. 2024. MMGNN: A Molecular Merged Graph Neural Network for Explain- able Solvation Free Energy Prediction. InProceedings of the 33rd International Joint Conference on Artificial Intelligence. 5808–5816

    work page 2024

  7. [7]

    Aniruddha Ganguly, Debolina Chatterjee, Wentao Huang, Jie Zhang, Alisa Yurovsky, Travis Steele Johnson, and Chao Chen. 2025. MERGE: Multi-faceted Hierarchical Graph-based GNN for Gene Expression Prediction from Whole Slide Histopathology Images. InProceedings of the Computer Vision and Pattern Recognition Conference. 15611–15620

    work page 2025

  8. [8]

    Jing Gao, Ce Zheng, Laszlo A Jeni, and Zackory Erickson. 2025. DiSRT-In-Bed: Diffusion-Based Sim-to-Real Transfer Framework for In-Bed Human Mesh Re- covery. InProceedings of the Computer Vision and Pattern Recognition Conference. 1829–1838

    work page 2025

  9. [9]

    Schoenholz, Patrick F

    Justin Gilmer, Samuel S. Schoenholz, Patrick F. Riley, Oriol Vinyals, and George E. Dahl. 2017. Neural Message Passing for Quantum Chemistry. InProceedings of the 34th International Conference on Machine Learning. 1263–1272

    work page 2017

  10. [10]

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. 2020. Generative Adversarial Networks.Commun. ACM63, 11 (2020), 139–144

    work page 2020

  11. [11]

    Yuqin He, Tengfei Ma, Chaoyi Li, Pengsen Ma, Hongxin Xiang, Jianmin Wang, Yip- ing Liu, Bosheng Song, and Xiangxiang Zeng. 2025. ImageDDI: Image-enhanced molecular motif sequence representation for drug-drug interaction prediction. Information Fusion(2025), 103574

    work page 2025

  12. [12]

    Weihua Hu, Bowen Liu, Joseph Gomes, Marinka Zitnik, Percy Liang, Vijay Pande, and Jure Leskovec. 2020. Strategies for Pre-Training Graph Neural Networks. In Proceedings of the 8th International Conference on Learning Representations

    work page 2020

  13. [13]

    ITCM. 2021. Integrated Traditional Chinese Medicine. http://itcm.biotcm.net/ index.html. Accessed: 2021

    work page 2021

  14. [14]

    Pengcheng Jiang, Cao Xiao, Tianfan Fu, Parminder Bhatia, Taha Kass-Hout, Jimeng Sun, and Jiawei Han. 2025. Bi-Level Contrastive Learning for Knowledge- Enhanced Molecule Representations. InProceedings of the 39th AAAI Conference on Artificial Intelligence. 352–360

    work page 2025

  15. [15]

    Joonyoung F Joung, Minhi Han, Minseok Jeong, and Sungnam Park. 2020. Exper- imental Database of Optical Properties of Organic Compounds.Scientific Data7, 1 (2020), 295

    work page 2020

  16. [16]

    Thomas N Kipf and Max Welling. 2017. Semi-Supervised Classification with Graph Convolutional Networks. InProceedings of the 5th International Conference on Learning Representations

    work page 2017

  17. [17]

    Na, Sungwon Kim, Junseok Lee, and Chanyoung Park

    Namkyeong Lee, Dongmin Hyun, Gyoung S. Na, Sungwon Kim, Junseok Lee, and Chanyoung Park. 2023. Conditional Graph Information Bottleneck for Molecular Relational Learning. InProceedings of the 40th International Conference on Machine Learning. 18852–18871

    work page 2023

  18. [18]

    Namkyeong Lee, Kanghoon Yoon, Gyoung S Na, Sein Kim, and Chanyoung Park

    work page

  19. [19]

    In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining

    Shift-Robust Molecular Relational Learning with Causal Substructure. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 1200–1212

    work page

  20. [20]

    Fengcheng Li, Minjie Mou, Xiaoyi Li, Weize Xu, Jiayi Yin, Yang Zhang, and Feng Zhu. 2024. DrugMAP 2.0: Molecular Atlas and Pharma-Information of All Drugs. Nucleic Acids Research53, D1 (09 2024), D1372–D1382

    work page 2024

  21. [21]

    Zhaoyang Li, Yuan Wang, Wangkai Li, Tianzhu Zhang, and Xiang Liu. 2025. Dual- Agent Optimization Framework for Cross-Domain Few-Shot Segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 9849–9859

    work page 2025

  22. [22]

    Zimeng Li, Shichao Zhu, Bin Shao, Xiangxiang Zeng, Tong Wang, and Tie- Yan Liu. 2023. DSN-DDI: An Accurate and Generalized Framework for Drug- Drug Interaction Prediction by Dual-View Representation Learning.Briefings in Bioinformatics24, 1 (2023), bbac597

    work page 2023

  23. [23]

    Qiujie Lv, Guanxing Chen, Ziduo Yang, Weihe Zhong, and Calvin Yu-Chian Chen. 2025. Meta-MolNet: A Cross-Domain Benchmark for Few Examples Drug Discovery.IEEE Transactions on Neural Networks and Learning Systems36, 3 (2025), 4849–4863

    work page 2025

  24. [24]

    Aleksandr V Marenich, Casey P Kelly, Jason D Thompson, Gregory D Hawkins, Candee C Chambers, David J Giesen, Paul Winget, Christopher J Cramer, and Donald G Truhlar. 2020. Minnesota Solvation Database (MNSOL) Version 2012. (2020)

    work page 2020

  25. [25]

    Arnold K Nyamabo, Hui Yu, and Jian-Yu Shi. 2021. SSI-DDI: Substructure- Substructure Interactions for Drug-Drug Interaction Prediction.Briefings in Bioinformatics22, 6 (2021), bbab133

    work page 2021

  26. [26]

    Yashaswi Pathak, Siddhartha Laghuvarapu, Sarvesh Mehta, and U Deva Priyaku- mar. 2020. Chemically Interpretable Graph Interaction Network for Prediction of Pharmacokinetic Properties of Drug-Like Molecules. InProceedings of the 34th AAAI Conference on Artificial Intelligence. 873–880

    work page 2020

  27. [27]

    Qucheng Peng, Ce Zheng, and Chen Chen. 2024. A Dual-Augmentor Framework for Domain Generalization in 3D Human Pose Estimation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2240–2249

    work page 2024

  28. [28]

    Zhixiang Ren and Huan Xu. 2024. TCM Modernization. https://www.tcmm.net. cn/zh-hans/. Accessed: 2024

    work page 2024

  29. [29]

    Zhonghao Ren, Xiangxiang Zeng, Yizhen Lao, Zhuhong You, Yifan Shang, Quan Zou, and Chen Lin. 2025. Predicting Rare Drug-Drug Interaction Events with Dual-Granular Structure-Adaptive and Pair Variational Representation.Nature Communications16, 1 (2025), 3997

    work page 2025

  30. [30]

    Nazanin Mohammadi Sepahvand, Eleni Triantafillou, Hugo Larochelle, Doina Precup, James J Clark, Daniel M Roy, and Gintare Karolina Dziugaite. 2025. Selective Unlearning via Representation Erasure Using Domain Adversarial Training. InProceedings of the the 13rd International Conference on Learning Representations

    work page 2025

  31. [31]

    Yao Tian, Jiacai Yi, Ningning Wang, Chengkun Wu, Jinfu Peng, Shao Liu, Guoping Yang, and Dongsheng Cao. 2024. DDInter 2.0: An Enhanced Drug Interaction Resource with Expanded Data Coverage, New Interaction Types, and Improved User Interface.Nucleic Acids Research53, D1 (08 2024), D1356–D1362

    work page 2024

  32. [32]

    Petar Veličković, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua Bengio. 2018. Graph Attention Networks. InProceedings of the 6th International Conference on Learning Representations

    work page 2018

  33. [33]

    Florence H Vermeire and William H Green. 2021. Transfer Learning for Solvation Free Energies: From Quantum Chemistry to Experiments.Chemical Engineering Journal418 (2021), 129307

    work page 2021

  34. [34]

    Ruijia Wang, Haoran Dai, Cheng Yang, Le Song, and Chuan Shi. 2024. Advancing Molecule Invariant Representation via Privileged Substructure Identification. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 3188–3199

    work page 2024

  35. [35]

    Yingheng Wang, Yaosen Min, Xin Chen, and Ji Wu. 2021. Multi-View Graph Contrastive Representation Learning for Drug-Drug Interaction Prediction. In Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Trovato et al. Proceedings of the 30th ACM on Web Conference. 2921–2933

    work page 2021

  36. [36]

    Zixu Wang, Yangyang Chen, Pengsen Ma, Zhou Yu, Jianmin Wang, Yuansheng Liu, Xiucai Ye, Tetsuya Sakurai, and Xiangxiang Zeng. 2025. Image-Based Gener- ation for Molecule Design with SketchMol.Nature Machine Intelligence(2025), 1–12

    work page 2025

  37. [37]

    Hongxin Xiang, Shuting Jin, Jun Xia, Man Zhou, Jianmin Wang, Li Zeng, and Xiangxiang Zeng. 2024. An Image-Enhanced Molecular Graph Representation Learning Framework. InProceedings of the 33rd International Joint Conference on Artificial Intelligence. 6107–6115

    work page 2024

  38. [38]

    Hongxin Xiang, Li Zeng, Linlin Hou, Kenli Li, Zhimin Fu, Yunguang Qiu, Ruth Nussinov, Jianying Hu, Michal Rosen-Zvi, Xiangxiang Zeng, et al . 2024. A Molecular Video-Derived Foundation Model for Scientific Drug Discovery.Nature Communications15, 1 (2024), 9696

    work page 2024

  39. [39]

    Gezheng Xu, Li Yi, Pengcheng Xu, Jiaqi Li, Ruizhi Pu, Changjian Shui, Charles Ling, A Ian McLeod, and Boyu Wang. 2025. Unraveling the Mysteries of Label Noise in Source-Free Domain Adaptation: Theory and Practice.IEEE Transactions on Pattern Analysis and Machine Intelligence(2025)

    work page 2025

  40. [40]

    Huali Xu, Shuaifeng Zhi, Shuzhou Sun, Vishal Patel, and Li Liu. 2025. Deep Learning for Cross-Domain Few-Shot Visual Recognition: A Survey.Comput. Surveys57, 8 (2025), 1–37

    work page 2025

  41. [41]

    Ziduo Yang, Weihe Zhong, Qiujie Lv, and Calvin Yu-Chian Chen. 2022. Learning Size-Adaptive Molecular Substructures for Explainable Drug-Drug Interaction Prediction by Substructure-Aware Graph Neural Network.Chemical Science13, 29 (2022), 8693–8703

    work page 2022

  42. [42]

    Xiangxiang Zeng, Hongxin Xiang, Linhui Yu, Jianmin Wang, Kenli Li, Ruth Nussi- nov, and Feixiong Cheng. 2022. Accurate prediction of molecular properties and drug targets using a self-supervised image representation learning framework. Nature Machine Intelligence4, 11 (2022), 1004–1016

    work page 2022

  43. [43]

    Peiliang Zhang, Jingling Yuan, Chao Che, Yongjun Zhu, and Lin Li. 2025. Subgraph Information Bottleneck with Causal Dependency for Stable Molecular Relational Learning. InProceedings of the 34th International Joint Conference on Artificial Intelligence

    work page 2025

  44. [44]

    Ran Zhang, Xuezhi Wang, Guannan Liu, Pengyang Wang, Yuanchun Zhou, and Pengfei Wang. 2025. Motif-Oriented Representation Learning with Topology Refinement for Drug-Drug Interaction Prediction. InProceedings of the 39th AAAI Conference on Artificial Intelligence. 1102–1110

    work page 2025

  45. [45]

    Shuai Zhang, Junfeng Fang, Xuqiang Li, ALAN XIA, Ye Wei, Wenjie Du, Yang Wang, et al. 2025. Iterative Substructure Extraction for Molecular Relational Learning with Interactive Graph Information Bottleneck. InProceedings of the 13rd International Conference on Learning Representations

    work page 2025

  46. [46]

    Wen Zhang, Yanlin Chen, Feng Liu, Fei Luo, Gang Tian, and Xiaohong Li. 2017. Predicting Potential Drug-Drug Interactions by Integrating Chemical, Biological, Phenotypic and Network Data.BMC Bioinformatics18 (2017), 1–12

    work page 2017

  47. [47]

    −log P𝑡 (𝑍 (𝑔) 𝑡 ) P𝑠 (𝑍 (𝑔) 𝑡 ) +P 𝑡 (𝑍 (𝑔) 𝑡 ) # +E𝑍 (𝑔) 𝑠 ∼P𝑠

    Yiheng Zhu, Mingyang Li, Junlong Liu, Kun Fu, Jiansheng Wu, Qiuyi Li, Mingze Yin, Jieping Ye, Jian Wu, and Zheng Wang. 2025. A Generalist Cross-Domain Molecular Learning Framework for Structure-Based Drug Discovery.arXiv preprint arXiv:2503.04362(2025). Cross-Domain Molecular Relational Learning: Leveraging Chemical Structure-Activity Analysis Conference ...

    work page arXiv 2025