Artykuł w czasopiśmie
Brak miniatury
Licencja
ClosedAccessDostęp zamknięty
 

Molecule Edit Graph Attention Network: Modeling Chemical Reactions as Sequences of Graph Edits

Uproszczony widok
dc.abstract.enThe central challenge in automated synthesis planning is to be able to generate and predict outcomes of a diverse set of chemical reactions. In particular, in many cases, the most likely synthesis pathway cannot be applied due to additional constraints, which requires proposing alternative chemical reactions. With this in mind, we present Molecule Edit Graph Attention Network (MEGAN), an end-to-end encoder–decoder neural model. MEGAN is inspired by models that express a chemical reaction as a sequence of graph edits, akin to the arrow pushing formalism. We extend this model to retrosynthesis prediction (predicting substrates given the product of a chemical reaction) and scale it up to large data sets. We argue that representing the reaction as a sequence of edits enables MEGAN to efficiently explore the space of plausible chemical reactions, maintaining the flexibility of modeling the reaction in an end-to-end fashion and achieving state-of-the-art accuracy in standard benchmarks. Code and trained models are made available online at https://github.com/molecule-one/megan.
dc.affiliationUniwersytet Warszawski
dc.contributor.authorSacha, Mikołaj
dc.contributor.authorBłaż, Mikołaj
dc.contributor.authorDąbrowski-Tumański, Paweł
dc.contributor.authorByrski, Piotr
dc.contributor.authorWłodarczyk-Pruszyński, Paweł
dc.contributor.authorJastrzębski, Stanisław
dc.contributor.authorLOSKA, Rafał
dc.contributor.authorChromiński, Mikołaj
dc.date.accessioned2024-01-25T12:53:38Z
dc.date.available2024-01-25T12:53:38Z
dc.date.issued2021
dc.description.financePublikacja bezkosztowa
dc.description.number7
dc.description.volume61
dc.identifier.doi10.1021/ACS.JCIM.1C00537
dc.identifier.issn1549-9596
dc.identifier.urihttps://repozytorium.uw.edu.pl//handle/item/112885
dc.identifier.weblinkhttps://pubs.acs.org/doi/pdf/10.1021/acs.jcim.1c00537
dc.languageeng
dc.pbn.affiliationchemical sciences
dc.relation.ispartofJournal of Chemical Information and Modeling
dc.relation.pages3273-3284
dc.rightsClosedAccess
dc.sciencecloudnosend
dc.titleMolecule Edit Graph Attention Network: Modeling Chemical Reactions as Sequences of Graph Edits
dc.typeJournalArticle
dspace.entity.typePublication