Abstract:Generative models provide an efficient way to analyze and understand unlabeled data, creating the latent space for data modeling and generation. Since the interpretation of latent space usually requires implicit expert knowledge, this human-centered feature makes visual analytic methods effective. In the filed of computer chemistry, some research have applied generative models to generate chemical spaces and generated new molecules by sampling in the latent space. However, the latent space is typically high and sparse, and there may be a large number of “dead zones”, which may lead to decoding sample points from the latent space are noisy or invalid. Therefore, it is extremely challenging to efficiently search and traverse the latent space and generate new molecules with the desired properties. This paper aims to propose a visualization system for interactive exploration of latent space, which inspires the researchers to design new potential molecules with desired properties. The main work of this paper is as follows: First, we investigate a series of literature on the application of generative networks to drug design and synthesis, and interview experts with computer chemistry background to summarize the requirements and tasks. Second, based on the above requirements and tasks, we propose a graph-based latent space traversal and interpolation algorithms and neighborhood sampling algorithms. This can improve the number of generated potential molecules and the speed of discovery of similar molecules. Then, we conduct comparison experiments to verify the effectiveness of the algorithms. Finally, we design visualization system and then conduct the case study and user study to verify the effectiveness of the visualization system.

Graph-based Latent Space Traversal for New Molecules Discovery.

Navigating Chemical Space with Latent Flows

Advances in Deep Learning-Based 3D Molecular Generative Models

Projecting Molecules into Synthesizable Chemical Spaces

Visualizing Deep Graph Generative Models for Drug Discovery

Target-aware Molecular Graph Generation

CELLS: Cost-Effective Evolution in Latent Space for Goal-Directed Molecular Generation.

Generative Artificial Intelligence for Navigating Synthesizable Chemical Space

Molecule Generation for Drug Design: a Graph Learning Perspective

Latent Chemical Space Searching for Plug-in Multi-objective Molecule Generation

Molecule Design by Latent Space Energy-Based Modeling and Gradual Distribution Shifting

Molecular Graph Generation via Geometric Scattering

Generative models for molecular discovery: Recent advances and challenges

Representing Molecules as Random Walks Over Interpretable Grammars

MoFlow: An Invertible Flow Model for Generating Molecular Graphs

Graph-Based Generative Adversarial Networks for Molecular Generation with Noise Diffusion.

3M-Diffusion: Latent Multi-Modal Diffusion for Language-Guided Molecular Structure Generation

SynthFormer: Equivariant Pharmacophore-based Generation of Molecules for Ligand-Based Drug Design

Innovative Drug-like Molecule Generation from Flow-based Generative Model

Navigating the frontier of drug-like chemical space with cutting-edge generative AI models

Assessing Methods and Obstacles in Chemical Space Exploration