AsteriX: a Web server to automatically extract ligand coordinates from figures in PDF articles.

Area of Science:

• Computational chemistry
• Cheminformatics
• Drug discovery

Background:

• Chemical structure coordinates are vital for drug design.
• Ligand coordinates are often found in scientific literature, requiring manual reconstruction.
• Manual reconstruction is a time-consuming process.

Purpose of the Study:

• To develop a web server, AsteriX, for reconstructing 3D ligand coordinates from 2D images in PDF files.
• To automate the extraction of ligand information from scientific literature.

Main Methods:

• AsteriX analyzes PDF images to identify potential ligand structures.
• It extracts connectivity and atom type information to reconstruct 3D coordinates.
• The server allows user augmentation and correction of partially reconstructed 3D structures.

Main Results:

• AsteriX identified 88% of 3249 ligand structures as chemical diagrams in test articles.
• Approximately 50% of identified diagrams were correctly interpreted as 3D structures.
• An additional third of structures required only minor manual corrections.

Conclusions:

• AsteriX automates the reconstruction of 3D ligand coordinates from 2D images, significantly reducing manual effort.
• The web server provides a valuable tool for drug design by facilitating access to ligand coordinates from literature.
• User-corrected and validated 3D reconstructions are made freely available in common drug design formats.