Scholarlink

<p>Methods for automatic chemical retrosynthesis have found recent success through the application of models traditionally built for natural language processing, primarily through transformer neural networks. These models have demonstrated significant ability to translate between the SMILES encodings of chemical products and reactants, but are constrained as a result of their autoregressive nature. We propose <InlineEquation ID="IEq3"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="13321_2025_1056_Article_IEq1.gif" Format="GIF" Height="13" Rendition="HTML" Resolution="72" Type="Linedraw" Width="54" /> </InlineMediaObject> <EquationSource Format="TEX">\(\texttt {DiffER}\)</EquationSource> <EquationSource Format="MATHML"><math> <mi mathvariant="monospace">DiffER</mi> </math></EquationSource> </InlineEquation>, an alternative template-free method for single-step retrosynthesis prediction in the form of categorical diffusion, which allows the entire output SMILES sequence to be predicted in unison. We construct an ensemble of diffusion models which achieves state-of-the-art performance for top-1 accuracy and competitive performance for top-3, top-5, and top-10 accuracy among template-free methods. We prove that <InlineEquation ID="IEq4"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="13321_2025_1056_Article_IEq1.gif" Format="GIF" Height="13" Rendition="HTML" Resolution="72" Type="Linedraw" Width="54" /> </InlineMediaObject> <EquationSource Format="TEX">\(\texttt {DiffER}\)</EquationSource> <EquationSource Format="MATHML"><math> <mi mathvariant="monospace">DiffER</mi> </math></EquationSource> </InlineEquation> is a strong baseline for a new class of template-free model and is capable of learning a variety of synthetic techniques used in laboratory settings.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

\(\texttt {DiffER}\): categorical diffusion ensembles for single-step chemical retrosynthesis

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月29日

查看全文
  • Sean Current,
  • Ziqi Chen,
  • Daniel Adu-Ampratwum,
  • Xia Ning,
  • Srinivasan Parthasarathy

摘要

Methods for automatic chemical retrosynthesis have found recent success through the application of models traditionally built for natural language processing, primarily through transformer neural networks. These models have demonstrated significant ability to translate between the SMILES encodings of chemical products and reactants, but are constrained as a result of their autoregressive nature. We propose \(\texttt {DiffER}\) DiffER , an alternative template-free method for single-step retrosynthesis prediction in the form of categorical diffusion, which allows the entire output SMILES sequence to be predicted in unison. We construct an ensemble of diffusion models which achieves state-of-the-art performance for top-1 accuracy and competitive performance for top-3, top-5, and top-10 accuracy among template-free methods. We prove that \(\texttt {DiffER}\) DiffER is a strong baseline for a new class of template-free model and is capable of learning a variety of synthetic techniques used in laboratory settings.