Pushing the Boundaries of Molecular Representation for Drug Discovery with the Graph Attention Mechanism

Abstract

Hunting for chemicals with favorable pharmacological, toxicological, and pharmacokinetic properties remains a formidable challenge for drug discovery. Deep learning provides us with powerful tools to build predictive models that are appropriate for the rising amounts of data, but the gap between what these neural networks learn and what human beings can comprehend is growing. Moreover, this gap may induce distrust and restrict deep learning applications in practice. Here, we introduce a new graph neural network architecture called Attentive FP for molecular representation that uses a graph attention mechanism to learn from relevant drug discovery data sets. We demonstrate that Attentive FP achieves state-of-the-art predictive performances on a variety of data sets and that what it learns is interpretable. The feature visualization for Attentive FP suggests that it automatically learns nonlocal intramolecular interactions from specified tasks, which can help us gain chemical insights directly from data beyond human perception.

Could Graph Neural Networks Learn Better Molecular Representation for Drug Discovery? A Comparison Study of Descriptor-based and Graph-based Models

“…Recently, a GNN method: Attentive FP, has gained increasing attention from the scientific community. [28] As shown in Table 1, Attentive FP yields the best predictions to 6 out of 11 benchmark datasets, including 2 regression tasks (ESOL and FreeSolv) and 4 classification tasks (MUV, BBBP, ToxCast and ClinTox), highlighting its impressive performance in modelling a variety of chemical properties in comparison with several other graph-based methods. A majority of those studies claimed that graph-based models are typically superior or comparable to traditional descriptor-based models, [25,[31][32][33][34][35][36] and only a few studies gave the opposite conclusions.…”

“…To well compare the performance of descriptor-based and graph-based models, the dataset collection related to drug discovery used by Attentive FP was also adopted in this study. [28] This dataset collection contains 11 different datasets originally reported in MoleculeNet for a variety of chemical endpoints. [33] In the study reported by Xiong et al, [28] the molecules that could not be successfully processed by RDKit [44] or the Attentive FP model were excluded from the original datasets.…”

“…[28] This dataset collection contains 11 different datasets originally reported in MoleculeNet for a variety of chemical endpoints. [33] In the study reported by Xiong et al, [28] the molecules that could not be successfully processed by RDKit [44] or the Attentive FP model were excluded from the original datasets. The details of those datasets are summarized in Table 2.…”

“…The state-of-the-art accuracy of GNN models in property prediction has been well represented. [18,25,[28][29][30][31][32][33] The representative GNN models and their statistical performances on the MoleculeNet benchmark datasets [33] are summarized in Table 1. As we can see, their performances on the benchmark datasets vary from one to another, which may be attributed to the discrepancies on the model architectures, evaluation methods, training strategies and so on.…”

DeepSnap-Deep Learning Approach Predicts Progesterone Receptor Antagonist Activity With High Performance

“…In the DL, the features within an image are extracted by a convolution process with CNNs. Therefore, by specifying the convolutional region(s) using combination analysis with other, new methods to visualize the region(s) of the feature(s) (Selvaraju et al, 2016;Xu et al, 2017;Farahat et al, 2019;Oh et al, 2019;Xiong et al, 2019), the important part or area of the chemical structure necessary for prediction model construction could be estimated. Secondly, the optimal 3Dstructuring rules have not been defined.…”