Department of Computer Science, Hunter College, The City University of New York, 695 Park Ave, New York, NY, 10065, USA.
The Graduate Center, The City University of New York, 356 5th Ave, New York, NY, 10016, USA.
BMC Bioinformatics. 2022 May 2;23(Suppl 3):158. doi: 10.1186/s12859-022-04681-3.
Drug discovery is time-consuming and costly. Machine learning, especially deep learning, shows great potential in quantitative structure-activity relationship (QSAR) modeling to accelerate drug discovery process and reduce its cost. A big challenge in developing robust and generalizable deep learning models for QSAR is the lack of a large amount of data with high-quality and balanced labels. To address this challenge, we developed a self-training method, Partially LAbeled Noisy Student (PLANS), and a novel self-supervised graph embedding, Graph-Isomorphism-Network Fingerprint (GINFP), for chemical compounds representations with substructure information using unlabeled data. The representations can be used for predicting chemical properties such as binding affinity, toxicity, and others. PLANS-GINFP allows us to exploit millions of unlabeled chemical compounds as well as labeled and partially labeled pharmacological data to improve the generalizability of neural network models.
We evaluated the performance of PLANS-GINFP for predicting Cytochrome P450 (CYP450) binding activity in a CYP450 dataset and chemical toxicity in the Tox21 dataset. The extensive benchmark studies demonstrated that PLANS-GINFP could significantly improve the performance in both cases by a large margin. Both PLANS-based self-training and GINFP-based self-supervised learning contribute to the performance improvement.
To better exploit chemical structures as an input for machine learning algorithms, we proposed a self-supervised graph neural network-based embedding method that can encode substructure information. Furthermore, we developed a model agnostic self-training method, PLANS, that can be applied to any deep learning architectures to improve prediction accuracies. PLANS provided a way to better utilize partially labeled and unlabeled data. Comprehensive benchmark studies demonstrated their potentials in predicting drug metabolism and toxicity profiles using sparse, noisy, and imbalanced data. PLANS-GINFP could serve as a general solution to improve the predictive modeling for QSAR modeling.
药物发现是一个耗时且昂贵的过程。机器学习，尤其是深度学习，在定量构效关系（QSAR）建模方面显示出了巨大的潜力，可以加速药物发现过程并降低成本。为 QSAR 开发稳健且可推广的深度学习模型的一个主要挑战是缺乏具有高质量和平衡标签的大量数据。为了解决这个挑战，我们开发了一种自训练方法，即部分标记有噪声的学生（PLANS），以及一种新的自监督图嵌入方法，即图同构网络指纹（GINFP），用于具有子结构信息的化学化合物表示，可以使用无标签数据来预测结合亲和力、毒性等化学性质。PLANS-GINFP 允许我们利用数百万个未标记的化学化合物以及标记和部分标记的药理学数据来提高神经网络模型的泛化能力。
我们评估了 PLANS-GINFP 在 CYP450 数据集和 Tox21 数据集的细胞色素 P450（CYP450）结合活性和化学毒性预测方面的性能。广泛的基准研究表明，PLANS-GINFP 可以在这两种情况下显著提高性能，且提高幅度很大。基于 PLANS 的自训练和基于 GINFP 的自监督学习都有助于提高性能。
为了更好地将化学结构作为机器学习算法的输入，我们提出了一种基于自监督图神经网络的嵌入方法，可以编码子结构信息。此外，我们开发了一种模型不可知的自训练方法 PLANS，可以应用于任何深度学习架构，以提高预测精度。PLANS 提供了一种更好地利用部分标记和未标记数据的方法。综合基准研究表明，它们在使用稀疏、嘈杂和不平衡的数据预测药物代谢和毒性特征方面具有潜力。PLANS-GINFP 可以作为一种通用解决方案，用于提高 QSAR 建模的预测建模能力。
