ABT-MPNN: an atom-bond transformer-based message-passing neural network for molecular property prediction

Simple item page
dc.contributor.authorLiu, Chengyou
dc.contributor.authorSun, Yan
dc.contributor.authorDavis, Rebecca
dc.contributor.authorCardona, Silvia T.
dc.contributor.authorHu, Pingzhao
dc.date.accessioned2023-03-02T17:02:25Z
dc.date.available2023-03-02T17:02:25Z
dc.date.issued2023-02-26
dc.date.updated2023-03-01T04:47:20Z
dc.description.abstractAbstract Graph convolutional neural networks (GCNs) have been repeatedly shown to have robust capacities for modeling graph data such as small molecules. Message-passing neural networks (MPNNs), a group of GCN variants that can learn and aggregate local information of molecules through iterative message-passing iterations, have exhibited advancements in molecular modeling and property prediction. Moreover, given the merits of Transformers in multiple artificial intelligence domains, it is desirable to combine the self-attention mechanism with MPNNs for better molecular representation. We propose an atom-bond transformer-based message-passing neural network (ABT-MPNN), to improve the molecular representation embedding process for molecular property predictions. By designing corresponding attention mechanisms in the message-passing and readout phases of the MPNN, our method provides a novel architecture that integrates molecular representations at the bond, atom and molecule levels in an end-to-end way. The experimental results across nine datasets show that the proposed ABT-MPNN outperforms or is comparable to the state-of-the-art baseline models in quantitative structure–property relationship tasks. We provide case examples of Mycobacterium tuberculosis growth inhibitors and demonstrate that our model's visualization modality of attention at the atomic level could be an insightful way to investigate molecular atoms or functional groups associated with desired biological properties. The new model provides an innovative way to investigate the effect of self-attention on chemical substructures and functional groups in molecular representation learning, which increases the interpretability of the traditional MPNN and can serve as a valuable way to investigate the mechanism of action of drugs.en_US
dc.identifier.citationJournal of Cheminformatics. 2023 Feb 26;15(1):29
dc.identifier.citationJournal of Cheminformatics. 2023 Feb 26;15(1):29
dc.identifier.urihttps://doi.org/10.1186/s13321-023-00698-9
dc.identifier.urihttp://hdl.handle.net/1993/37187
dc.language.isoengen_US
dc.language.rfc3066en
dc.publisherBMCen_US
dc.rightsopen accessen_US
dc.rights.holderThe Author(s)
dc.subjectMessage-passing neural networksen_US
dc.subjectAttention mechanismen_US
dc.subjectMolecular representationsen_US
dc.subjectAtom-bond Transformer message-passing neural networken_US
dc.subjectMolecular property predictionen_US
dc.subjectBiological activity predictionen_US
dc.titleABT-MPNN: an atom-bond transformer-based message-passing neural network for molecular property predictionen_US
dc.typeresearch articleen_US
local.author.affiliationRady Faculty of Health Sciences::Max Rady College of Medicine::Department of Biochemistry and Medical Geneticsen_US
oaire.citation.issue1en_US
oaire.citation.startPage29en_US
oaire.citation.titleJournal of Cheminformaticsen_US
oaire.citation.volume15en_US
project.funder.identifierhttps://doi.org/10.13039/501100000024en_US
project.funder.nameCanadian Institutes of Health Researchen_US
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
13321_2023_Article_698.pdf
Size:
3.41 MB
Format:
Adobe Portable Document Format
Description:
Download
License bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
license.txt
Size:
2.24 KB
Format:
Item-specific license agreed to upon submission
Description:
Download
Collections
University of Manitoba Scholarship