Chemical curation to improve data accuracy: recent development of the 3DMET database

Area of Science:

• Chemistry
• Biochemistry
• Computational Chemistry

Background:

• The three-dimensional structure database of natural metabolites (3DMET) was initially populated using auto-generated content from 2D chemical structures.
• Manual curation efforts began in 2009, sourcing new compounds from scientific literature.

Purpose of the Study:

• To detail the experiences and challenges encountered during the manual curation of the 3DMET database.
• To highlight the importance of accurate stereochemistry in 3D structure data for chemical databases.

Main Methods:

• Manual curation of natural metabolite structures from published works.
• Addressing issues related to digitizing 3D structures from document drawings, including stereochemistry.
• Focusing on the accurate representation of 3D configurations for chiral atoms.

Main Results:

• Accumulation of problems in digitizing 3D structures from document drawings during manual curation.
• Identification of stereochemistry as a critical factor in both auto-generation and manual digitization processes.
• Confirmation that manual curation remains essential for precise database entries, particularly for chiral compounds.

Conclusions:

• Manual curation is indispensable for ensuring the accuracy of 3D chemical structure databases, especially concerning stereochemistry.
• The experiences shared can aid researchers and developers in creating robust chemical structure database systems.
• Accurate 3D-configuration data, particularly for chiral centers in natural products, requires meticulous manual verification.