Login / Signup

Simulation-Based Algorithm for Two-Dimensional Chemical Structure Diagram Generation of Complex Molecules and Ligand-Protein Interactions.

Tomasz Frączek
Published in: Journal of chemical information and modeling (2016)
Computer programs for structure diagram generation (SDG) are indispensable cheminformatic tools that translate one- or three-dimensional (1D or 3D) chemical structure data stored in electronic formats to human-readable 2D depictions. Although many such programs are known, only a moderate part of chemical space can be handled by existing algorithms. For many classes of natural and synthetic compounds the results obtained with current SDG methods are illegible. A new algorithm based solely on a physical simulation of a molecule has been developed. The method allows for a general and global approach to avoid overlapping atoms and substituents. While the algorithm shows no advantage for trivial molecules, it shows superior performance over existing methods in depicting the most challenging compounds. The algorithm can generate chemically correct and legible 2D structure diagrams of many classes of natural and synthetic compounds that are intractable with existing SDG algorithms. The use of the method for generating schematic ligand-receptor interaction diagrams is also discussed.
Keyphrases
  • machine learning
  • deep learning
  • big data
  • artificial intelligence
  • public health
  • endothelial cells
  • high intensity
  • mental health
  • electronic health record