Versatile Molecular Editing via Multimodal and Group-optimized Generative Learning

Jun Zhang,Xiaohan Lin,Yijie Xia,Yupeng Huang,Shuo Liu,Yi Qin Gao

DOI: https://doi.org/10.26434/chemrxiv-2023-j2n6l-v2

2024-06-18

Abstract:Generating molecules with specific constituents and structures that exhibit desired properties is a crucial yet challenging task in the computer-aided design of functional molecules. This challenge arises from the discrete nature of the vast design space of molecules, which is subject to additional physical constraints such as symmetries. Exploration and optimization within this constrained discrete space pose difficulties for most machine learning methods. We approach these challenges from a novel multimodal perspective, and propose a multimodal generative method, MolEdit, to explicitly model both the discrete atomic constituents and their continuous positions in 3D space. MolEdit is based on a group-optimized score matching algorithm which preserves the symmetry of atomic positions on top of a joint model which generates the molecular constituents. Therefore, MolEdit simultaneously resolves the discrete and continuous optimization problems encountered in molecular generation. As a compelling feature, MolEdit can flexibly generate precisely edited molecular graphs subjected to various prompts which specify conditional information about molecular constituents as well as substructures. Overall, MolEdit not only advances the frontier of prompt-based molecular design, but also provides a template for transacting powerful modern generative methods for domain-specific scientific data in a multimodal way

Chemistry

What problem does this paper attempt to address?

The main focus of this paper is on a key but challenging problem in computer-aided design of functional molecules, which is molecular editing. Existing methods for molecular generation face optimization challenges in exploring the discrete design space, as this space is constrained by physical limitations such as molecular symmetry. The researchers propose a multimodal generation method called MolEdit, which models both the discrete atomic components of a molecule and their continuous positions in 3D space. MolEdit is based on a score matching algorithm using group optimization, which can generate molecule components while preserving atomic position symmetry. This approach addresses both discrete and continuous optimization problems in molecular generation and can flexibly generate precisely edited molecular graphs under various conditions. MolEdit not only advances the forefront of hint-based molecular design but also provides multimodal application templates for specific scientific data domains. In addition, the paper discusses quality evaluation, symmetry preservation, and considerations for physical information in molecular editing. Through its multimodal generation strategy and group optimization score matching learning, MolEdit is capable of handling both the discrete and continuous characteristics of molecules while addressing the issue of symmetry, resulting in molecular structures that better conform to physical laws.

Versatile Molecular Editing via Multimodal and Group-optimized Generative Learning

PrefixMol: Target- and Chemistry-aware Molecule Design Via Prefix Embedding

Domain-Agnostic Molecular Generation with Chemical Feedback

Multi-granularity Score-based Generative Framework Enables Efficient Inverse Design of Complex Organics

An Equivariant Generative Framework for Molecular Graph-Structure Co-Design

Structure-Based Drug Design via 3D Molecular Generative Pre-training and Sampling

Sculpting Molecules in Text-3D Space: A Flexible Substructure Aware Framework for Text-Oriented Molecular Optimization

Instruction Multi-Constraint Molecular Generation Using a Teacher-Student Large Language Model

Large Language Models as Molecular Design Engines

Chemical Language Model Linker: blending text and molecules with modular adapters

Generative Organic Electronic Molecular Design Informed by Quantum Chemistry

DecompOpt: Controllable and Decomposed Diffusion Models for Structure-based Molecular Optimization

Unlocking comprehensive molecular design across all scenarios with large language model and unordered chemical language

Machine learning-aided generative molecular design

Human-level molecular optimization driven by mol-gene evolution

High-throughput property-driven generative design of functional organic molecules

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

Structure-Based Molecule Optimization via Gradient-Guided Bayesian Update

Data-Efficient Graph Grammar Learning for Molecular Generation

GexMolGen: cross-modal generation of hit-like molecules via large language model encoding of gene expression signatures

ChatMol: Interactive Molecular Discovery with Natural Language