Universal J-Coupling Prediction

详细信息    查看全文

• 作者：Juuso Lehtivarjo ; Matthias Niemitz ; Samuli-Petrus Korhonen
• 刊名：Journal of Chemical Information and Modeling
• 出版年：2014
• 出版时间：March 24, 2014
• 年：2014
• 卷：54
• 期：3
• 页码：810-817
• 全文大小：360K
• 年卷期：v.54,no.3(March 24, 2014)
• ISSN：1549-960X

文摘

A data driven approach for small molecule J-coupling prediction is presented. The method is targeted for use as part of an automatic spectrum analysis, therefore emphasizing prediction coverage, maintainability, and speed in the design. The database search involves encoding the coupling path atom types into hash codes, which are used to retrieve the matching coupling constant entries from the database. The fast hash dictionary search is followed by a k Nearest Neighbors regression to resolve the substituent and conformational dependencies, parametrized with atomic charges, torsion angles, and steric bulk.