Data enhancement method for drugs under graph-structured representation

doi:10.11772/j.issn.1001-9081.2022040489

Journal of Computer Applications ›› 2023, Vol. 43 ›› Issue (4): 1136-1141.DOI: 10.11772/j.issn.1001-9081.2022040489

Special Issue: 数据科学与技术

• Data science and technology • Previous Articles Next Articles

Data enhancement method for drugs under graph-structured representation

Yinjiang CAI¹^,², Guangjun XU³, Xibo MA¹^,²()

^1.National Laboratory of Pattern Recognition （Institute of Automation，Chinese Academy of Sciences），Beijing 100190，China
^2.School of Artificial Intelligence，University of Chinese Academy of Sciences，Beijing 100049，China
^3.Data Center Beijing Branch，Agricultural Bank of China，Beijing 100088，China

Received:2022-04-14 Revised:2022-06-24 Accepted:2022-07-05 Online:2023-04-11 Published:2023-04-10
Contact: Xibo MA
About author:CAI Yinjiang， born in 1997， M. S. candidate. His research interests include intelligent medicine.
XU Guangjun， born in 1980， M. S.， engineer. His research interests include big data mining.
Supported by:
Natural Science Foundation of China(82090051);Excellent Member Program of Youth Innovation Promotion Association of Chinese Academy of Sciences(Y201930)

图结构表示下的药物数据增强方法

蔡引江¹^,², 许光俊³, 马喜波¹^,²()

^1.模式识别国家重点实验室(中国科学院自动化研究所), 北京 100190
^2.中国科学院大学人工智能学院, 北京 100049
^3.中国农业银行数据中心北京分部, 北京 100088

通讯作者: 马喜波
作者简介:蔡引江（1997—），男，江苏扬州人，硕士研究生，主要研究方向：智能医药；
许光俊（1980—），男，山东济南人，工程师，硕士，主要研究方向：大数据挖掘；
基金资助:
国家自然科学基金资助项目(82090051);中国科学院青年创新促进会优秀会员项目(Y201930)

Abstract

Abstract:

Small sample data can lead to over-fitting problems in machine learning models. In the field of drug development， most data tend to be small samples， which greatly limits the application of machine learning techniques in this field. To solve the above problem， a drug data enhancement method based on graph structure was proposed. The samples were perturbed by the proposed method and new similar samples were generated to expand the dataset. The proposed method are consisted of four sub-methods， which are node discarding method based on molecular backbone， edge discarding method based on molecular backbone， multi-sample splicing methods and hybrid strategy method. In specific， the perturbation of drug molecules was completed by the node and edge discarding method based on molecular backbone in the way of a small number of deletion operation on the composition and structure of drug molecules； the perturbation was completed by the multi-sample splicing method through using an addition operation to combine different molecules； in the hybrid strategy method， the diversity of data enhancement results was improved by combining the deletion and addition operation in a certain ratio. The proposed method improved the Area Under receiver operating characteristic Curve （AUC） of the drug attribute prediction baseline model MG-BERT （Molecular Graph Bidirectional Encoder Representations from Transformer） by 1.94% to 12.49% on public datasets BACE， BBBP， ToxCast and ClinTox. Experimental results demonstrate the effectiveness of the proposed method on small sample drug data enhancement.

Key words: small sample data, drug molecule, data enhancement, graph-structured representation, drug attribute prediction

摘要：

小样本数据会导致机器学习模型出现过拟合问题，而药物研发中的数据往往都具有小样本特性，这极大地限制了机器学习技术在该领域的应用。针对上述问题，提出了图结构下的药物数据增强方法。所提方法通过对样本微扰生成新的相似样本，以扩充数据集。所提方法包含4个子方法，分别是：基于分子骨干的节点丢弃法、基于分子骨干的边丢弃法、多样本拼接法以及混合策略法。其中，基于分子骨干的节点丢弃和边丢弃法通过少量删减药物分子的组成与结构完成对药物分子的微扰；多样本拼接法则使用一种增添性操作，通过组合不同分子完成微扰；混合策略法按一定比例配比删减性和增添性操作提升数据增强结果的多样性。在公开数据集BACE、BBBP、ToxCast和ClinTox上，所提方法为药物属性预测基线模型MG-BERT（Molecular Graph Bidirectional Encoder Representations from Transformer）的受试者工作特征曲线下面积（AUC）带来了1.94%~12.49%的提升。实验结果验证了所提方法在小样本药物数据增强上的有效性。

关键词: 小样本数据, 药物分子, 数据增强, 图结构表示, 药物属性预测

CLC Number:

TP181

Yinjiang CAI, Guangjun XU, Xibo MA. Data enhancement method for drugs under graph-structured representation[J]. Journal of Computer Applications, 2023, 43(4): 1136-1141.

蔡引江, 许光俊, 马喜波. 图结构表示下的药物数据增强方法[J]. 《计算机应用》唯一官方网站, 2023, 43(4): 1136-1141.

Figures/Tables 12

References 24

1	BAGHERIAN M， SABETI E， WANG K， et al. Machine learning approaches and databases for prediction of drug-target interaction： a survey paper［J］. Briefings in Bioinformatics， 2021， 22（1）： 247-269. 10.1093/bib/bbz157
2	XIONG J C， XIONG Z P， CHEN K X， et al. Graph neural networks for automated de novo drug design［J］. Drug Discovery Today， 2021， 26（6）： 1382-1393. 10.1016/j.drudis.2021.02.011
3	LIU R Q， WEI L， ZHANG P. A deep learning framework for drug repurposing via emulating clinical trials on real-world patient data［J］. Nature Machine Intelligence， 2021， 3（1）： 68-75. 10.1038/s42256-020-00276-w
4	TIAN Y J， ZHANG Y Q. A comprehensive survey on regularization strategies in machine learning［J］. Information Fusion， 2022， 80： 146-166. 10.1016/j.inffus.2021.11.005
5	SIMONYAN K， ZISSERMAN A. Very deep convolutional networks for large-scale image recognition［EB/OL］. （2015-04-10）［2022-02-20］..
6	HE K M， ZHANG X Y， REN S Q， et al. Deep residual learning for image recognition［C］// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2016： 770-778. 10.1109/cvpr.2016.90
7	DING K Z， XU Z， TONG H H， et al. Data augmentation for deep graph learning： a survey［J］. ACM SIGKDD Explorations Newsletter， 2022， 24（3）： 61-77. 10.1145/3575637.3575646
8	LIU S T， YING R， DONG H Z， et al. Local augmentation for graph neural networks［C］// Proceedings of the 39th International Conference on Machine Learning. New York： JMLR.org， 2022： 14054-14072.
9	VELIČKOVIĆ P， FEDUS W， HAMILTON W L， et al. Deep Graph Infomax［EB/OL］. （2018-12-21）［2022-07-05］..
10	FENG F L， HE X N， TANG J， et al. Graph adversarial training： dynamically regularizing based on graph structure［J］. IEEE Transactions on Knowledge and Data Engineering， 2021， 33（6）： 2493-2504. 10.1109/tkde.2019.2957786
11	XU Z， DU B X， TANG H H. Graph sanitation with application to node classification［C］// Proceedings of the 2022 ACM Web Conference. New York： ACM， 2022：1136-1147. 10.1145/3485447.3512180
12	ZHENG C， ZONG B， CHENG W， et al. Robust graph representation learning via neural sparsification［C］// Proceedings of the 37th International Conference on Machine Learning. New York： JMLR.org， 2020： 11458-11468.
13	ZHAO T X， ZHANG X， WANG S H. GraphSMOTE： imbalanced node classification on graphs with graph neural networks［C］// Proceedings of the 14th ACM International Conference on Web Search and Data Mining. New York： ACM， 2021： 833-841. 10.1145/3437963.3441720
14	KONDOR R I， LAFFERTY J. Diffusion kernels on graphs and other discrete structures［C］// Proceedings of the 19th International Conference on Machine Learning. San Francisco： Morgan Kaufmann Publishers Inc.， 2002： 315-322.
15	JIAO Y Z， XIONG Y， ZHANG J W， et al. Sub-graph contrast for scalable self-supervised graph representation learning［C］// Proceedings of the 2020 IEEE International Conference on Data Mining. Piscataway： IEEE， 2020： 222-231. 10.1109/icdm50108.2020.00031
16	QIU J Z， CHEN Q B， DONG Y X， et al. GCC： graph contrastive coding for graph neural network pre-training［C］// Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York： ACM， 2020： 1150-1160. 10.1145/3394486.3403168
17	YOU J X， YING R， REN X， et al. GraphRNN： generating realistic graphs with deep auto-regressive models［C］// Proceedings of the 35th International Conference on Machine Learning. New York： JMLR.org， 2018： 5708-5717.
18	BOJCHEVSKI A， SHCHUR O， ZÜGNER D， et al. NetGAN： generating graphs via random walks［C］// Proceedings of the 35th International Conference on Machine Learning. New York： JMLR.org， 2018： 610-619.
19	ZHANG H Y， CISSE M， DAUPHIN Y N， et al. mixup： beyond empirical risk minimization［EB/OL］. （2018-04-27）［2022-07-05］..
20	ZHAO T， LIU Y Z， NEVES L， et al. Data augmentation for graph neural networks［C］// Proceedings of the 35th AAAI Conference on Artificial Intelligence. Palo Alto， CA： AAAI Press， 2021： 11015-11023. 10.1609/aaai.v35i12.17315
21	ZHAO T， LIU G， WANG D H， et al. Counterfactual graph learning for link prediction［EB/OL］. （2022-06-03）［2022-07-05］..
22	曹一珉，蔡磊，高敬阳. 基于生成对抗网络的基因数据生成方法［J］. 计算机应用， 2022， 42（3）： 783-790.
	CAO Y M， CAI L， GAO J Y. Gene data generation method based on generative adversarial network［J］. Journal of Computer Applications， 2022， 42（3）： 783-790.
23	WU Z Q， RAMSUNDAR B， FEINBERG E N， et al. MoleculeNet： a benchmark for molecular machine learning［J］. Chemical Science， 2018， 9（2）： 513-530. 10.1039/c7sc02664a
24	ZHANG X C， WU C K， YANG Z J， et al. MG-BERT： leveraging unsupervised atomic representation learning for molecular property prediction［J］. Briefings in Bioinformatics， 2021， 22（6）： No.bbab152. 10.1093/bib/bbab152

数据集	化合物数	任务数	切分方式
BACE	1 522	1	随机分割
BBBP	2 053	1	随机分割
Tox21	8 014	12	Scaffold分割
ToxCast	8 615	617	Scaffold分割
ClinTox	1 478	2	随机分割

数据集	化合物数	任务数	切分方式
BACE	1 522	1	随机分割
BBBP	2 053	1	随机分割
Tox21	8 014	12	Scaffold分割
ToxCast	8 615	617	Scaffold分割
ClinTox	1 478	2	随机分割

方法	BACE	BBBP	Tox21	ToxCast	ClinTox
基线	0.809 1	0.878 2	0.827 0	0.731 3	0.698 1
方法1	0.857 5	0.875 1	0.826 4	0.741 2	0.722 5
方法2	0.835 3	0.869 4	0.827 2	0.755 2	0.696 9
方法3	0.852 2	0.882 8	0.820 5	0.777 4	0.780 1
方法4	0.862 1	0.895 2	0.826 8	0.767 9	0.785 3

方法	BACE	BBBP	Tox21	ToxCast	ClinTox
基线	0.809 1	0.878 2	0.827 0	0.731 3	0.698 1
方法1	0.857 5	0.875 1	0.826 4	0.741 2	0.722 5
方法2	0.835 3	0.869 4	0.827 2	0.755 2	0.696 9
方法3	0.852 2	0.882 8	0.820 5	0.777 4	0.780 1
方法4	0.862 1	0.895 2	0.826 8	0.767 9	0.785 3

方法	BACE	BBBP	Tox21	ToxCast	ClinTox
基线	0.734 5	0.938 3	0.396 4	0.090 4	0.610 3
方法1	0.810 3	0.950 2	0.382 5	0.111 4	0.663 8
方法2	0.958 7	0.812 2	0.385 1	0.109 6	0.682 2
方法3	0.800 0	0.949 5	0.385 0	0.114 4	0.695 8
方法4	0.831 4	0.952 8	0.393 2	0.128 7	0.663 9

Data enhancement method for drugs under graph-structured representation

图结构表示下的药物数据增强方法

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 12

References 24

Related Articles 6

Recommended Articles

Metrics

方法	BACE	BBBP	Tox21	ToxCast	ClinTox
基线	0.676 4	0.888 7	0.340 8	0.003 1	0.528 2
方法1	0.739 7	0.903 6	0.367 0	0.063 7	0.597 3
方法2	0.904 9	0.761 0	0.380 9	0.063 2	0.624 3
方法3	0.753 5	0.899 6	0.344 5	0.031 7	0.615 2
方法4	0.781 8	0.907 0	0.367 9	0.074 2	0.592 9

增强样本倍数	BACE	BBBP	Tox21	ToxCast	ClinTox
2	0.833 4	0.886 4	0.820 1	0.751 3	0.745 4
3	0.848 7	0.886 2	0.820 3	0.755 4	0.787 6
4	0.849 7	0.882 3	0.821 1	0.760 5	0.774 8
5	0.846 0	0.882 1	0.821 6	0.767 4	0.783 5
6	0.851 9	0.895 2	0.826 8	0.767 9	0.785 3
7	0.854 7	0.884 9	0.820 3	0.824 4	0.777 5
8	0.863 2	0.888 9	0.820 6	0.770 0	0.761 9

[1]	Ying YANG, Xiaoyan HAO, Dan YU, Yao MA, Yongle CHEN. Graph data generation approach for graph neural network model extraction attacks [J]. Journal of Computer Applications, 2024, 44(8): 2483-2492.
[2]	CHEN Guihui, YI Xin, LI Zhongbing, QIAN Jiren, CHEN Wu. Third-party construction target detection in aerial images of pipeline inspection based on improved YOLOv2 and transfer learning [J]. Journal of Computer Applications, 2020, 40(4): 1062-1068.
[3]	LIU Ziyan, WAN Peipei. Pedestrian re-identification feature extraction method based on attention mechanism [J]. Journal of Computer Applications, 2020, 40(3): 672-676.
[4]	FEI Dasheng, SONG Huihui, ZHANG Kaihua. Multi-level feature enhancement for real-time visual tracking [J]. Journal of Computer Applications, 2020, 40(11): 3300-3305.
[5]	YU Ying, WANG Lewei, ZHANG Yinglong. Data enhancement algorithm based on feature extraction preference and background color correlation [J]. Journal of Computer Applications, 2019, 39(11): 3172-3177.
[6]	WANG Keli, YUAN Hongchun. Aquatic animal image classification method based on transfer learning [J]. Journal of Computer Applications, 2018, 38(5): 1304-1308.