Molecular analysis and design using multimodal generative artificial intelligence via multi-agent modeling

Markus Buehler,Isabella Stewart

DOI: https://doi.org/10.26434/chemrxiv-2024-nwm7n

2024-04-16

Abstract:We report the use of a multimodal generative artificial intelligence framework, the X-LoRA-Gemma large language model (LLM), to analyze, design and test molecular design. The X-LoRA-Gemma model, inspired by biological principles and featuring ~7 billion parameters, dynamically reconfigures its structure through a dual-pass inference strategy to enhance its problem-solving abilities across diverse scientific domains. The model is used to first identify molecular engineering targets through a systematic human-AI and AI-AI self-driving multi-agent approach to elucidate key targets for molecular optimization to improve interactions between molecules. Next, a multi-agent generative design process is used that includes rational steps, reasoning and autonomous knowledge extraction. Target properties of the molecule are identified either using a Principal Component Analysis (PCA) of key molecular properties or sampling from the distribution of known molecular properties. The model is then used to generate a large set of candidate molecules, which are analyzed via their molecular structure, charge distribution, and other features. We validate that as predicted, increased dipole moment and polarizability is indeed achieved in the designed molecules. We anticipate an increasing integration of these techniques into the molecular engineering workflow, ultimately enabling the development of innovative solutions to address a wide range of societal challenges. We conclude with a critical discussion of challenges and opportunities of the use of multimodal generative AI for molecular engineering, analysis and design.

Chemistry

What problem does this paper attempt to address?

The paper aims to address key issues in molecular design and analysis, particularly by leveraging multimodal generative artificial intelligence techniques to optimize molecular properties. Specifically, the researchers have developed a large-scale language model (LLM) named X-LoRA-Gemma, which has 7 billion parameters and is inspired by biological principles. By dynamically reorganizing its structure to enhance problem-solving capabilities across different scientific domains, this model can identify molecular engineering targets, design new molecules, and validate their performance.

The core objective of the research is to demonstrate the application potential of the X-LoRA-Gemma model in the field of molecular design, especially in tasks such as quantum mechanical property prediction and molecular design. To achieve this goal, the researchers employed a series of methods:

1. **Identification of Molecular Design Targets**: Key molecular optimization targets are identified through a self-driven multi-agent approach between human-machine collaboration and AI.
2. **Multi-Agent Generative Design Process**: This includes rational steps, reasoning, and autonomous knowledge extraction to generate candidate molecules.
3. **Identification of Molecular Properties**: The properties of target molecules can be determined through principal component analysis (PCA) or by sampling from the distribution of known molecular properties.
4. **Molecular Generation and Analysis**: After the model generates a large number of candidate molecules, their molecular structures, charge distributions, and other features are analyzed.
5. **Performance Validation**: The designed molecules are experimentally validated to ensure they possess the expected increased properties such as dipole moment and polarizability.

Additionally, the research explores multi-agent interactions between the X-LoRA-Gemma model and other models to automate molecular design tasks, such as exploring molecular designs for organic electronic devices through self-driven interactions between two agents.

In summary, the goal of this paper is to demonstrate how the X-LoRA-Gemma model can be effectively applied to the molecular design process to advance fields such as new materials and drug discovery, and to address sustainability challenges.