Spherical GTM: A New Proposition for Visualization of Chemical Data

Area of Science:

• Computational chemistry
• Data visualization
• Cheminformatics

Background:

• Traditional Generative Topographic Mapping (GTM) uses a flat Euclidean space, which can lead to inaccuracies with non-flat data topologies.
• Visualizing complex chemical data, such as electron densities and molecular libraries, requires robust and accurate methods.

Purpose of the Study:

• To introduce and mathematically formalize the Spherical Generative Topographic Mapping (SGTM) method.
• To demonstrate SGTM's capability in visualizing chemical data, including electron densities and molecular libraries.
• To evaluate SGTM's accuracy compared to existing visualization techniques.

Main Methods:

• Development of the mathematical formalism for SGTM, incorporating a spherical manifold.
• Application of SGTM to visualize 2D electron density patterns of water and benzene.
• Utilization of SGTM for visualizing the CosMoPoly chemical library.

Main Results:

• SGTM successfully visualizes 2D electron density patterns and the CosMoPoly chemical library.
• Comparison with established methods shows SGTM provides improved accuracy in chemical structure depiction.
• The spherical manifold effectively addresses non-flat topology issues inherent in chemical data.

Conclusions:

• SGTM is a novel and accurate 3D data visualization method for chemical data.
• The use of a spherical manifold enhances the topological representation of chemical structures.
• SGTM offers a valuable tool for cheminformatics and computational chemistry research.