Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

PolyAgent: Large Language Model Agent for Polymer Design

Vani Nigam, Achuth Chandrasekhar, Amir Barati Farimani

TL;DR

PolyAgent tackles the bottleneck in polymer design by deploying a closed-loop, LLM-driven agent that orchestrates generation and property prediction through the Model Context Protocol (MCP). It combines a reaction-aware SMILES generator (Molecule Chef fine-tuned on the Open Macromolecule Genome dataset) with a Transformer-based property predictor (TransPolymer) to steer toward synthetically accessible, target-property polymers. The framework provides a terminal-based interface for property-targeted structure generation, guided by Synthetic Accessibility ($SA$) and Synthetic Complexity ($SC$) scores, with validation via the predictor server and comparisons to open resources like Polymer Genome. By demonstrating end-to-end generation, refinement, and validation, PolyAgent highlights a path toward more accessible, efficient polymer discovery, while outlining open challenges in data, evaluation, and tool integration for future work.

Abstract

On-demand Polymer discovery is essential for various industries, ranging from biomedical to reinforcement materials. Experiments with polymers have a long trial-and-error process, leading to long procedures and extensive resources. For these processes, machine learning has accelerated scientific discovery at the property prediction and latent space search fronts. However, laboratory researchers cannot readily access codes and these models to extract individual structures and properties due to infrastructure limitations. We present a closed-loop polymer structure-property predictor integrated in a terminal for early-stage polymer discovery. The framework is powered by LLM reasoning to provide users with property prediction, property-guided polymer structure generation, and structure modification capabilities. The SMILES sequences are guided by the synthetic accessibility score and the synthetic complexity score (SC Score) to ensure that polymer generation is as close as possible to synthetically accessible monomer-level structures. This framework addresses the challenge of generating novel polymer structures for laboratory researchers, thereby providing computational insights into polymer research.

PolyAgent: Large Language Model Agent for Polymer Design

TL;DR

PolyAgent tackles the bottleneck in polymer design by deploying a closed-loop, LLM-driven agent that orchestrates generation and property prediction through the Model Context Protocol (MCP). It combines a reaction-aware SMILES generator (Molecule Chef fine-tuned on the Open Macromolecule Genome dataset) with a Transformer-based property predictor (TransPolymer) to steer toward synthetically accessible, target-property polymers. The framework provides a terminal-based interface for property-targeted structure generation, guided by Synthetic Accessibility () and Synthetic Complexity () scores, with validation via the predictor server and comparisons to open resources like Polymer Genome. By demonstrating end-to-end generation, refinement, and validation, PolyAgent highlights a path toward more accessible, efficient polymer discovery, while outlining open challenges in data, evaluation, and tool integration for future work.

Abstract

On-demand Polymer discovery is essential for various industries, ranging from biomedical to reinforcement materials. Experiments with polymers have a long trial-and-error process, leading to long procedures and extensive resources. For these processes, machine learning has accelerated scientific discovery at the property prediction and latent space search fronts. However, laboratory researchers cannot readily access codes and these models to extract individual structures and properties due to infrastructure limitations. We present a closed-loop polymer structure-property predictor integrated in a terminal for early-stage polymer discovery. The framework is powered by LLM reasoning to provide users with property prediction, property-guided polymer structure generation, and structure modification capabilities. The SMILES sequences are guided by the synthetic accessibility score and the synthetic complexity score (SC Score) to ensure that polymer generation is as close as possible to synthetically accessible monomer-level structures. This framework addresses the challenge of generating novel polymer structures for laboratory researchers, thereby providing computational insights into polymer research.
Paper Structure (21 sections, 10 figures, 5 tables)

This paper contains 21 sections, 10 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Methods
  3. Polymer Informatics
  4. Generative Models
  5. Model Context Protocol (MCP)
  6. Agent Design
  7. Task Description
  8. Workflow
  9. Results
  10. Closed loop generation and prediction
  11. Generative Server
  12. Property-Aware Molecular Refinement
  13. Property Predictor Server
  14. Discussion
  15. Open Challenges and Future Work
  16. ...and 6 more sections

Figures (10)

  • Figure 1: Workflow illustrating how each tool is sent an input and how each of the tools generates an output in the PolyAgent
  • Figure 2: Tools available for usage in the MCP servers. The figure also shows how the tools access their respective utility and deliver it to the users.
  • Figure 3: Workflow of the terminal responses (blue: missteps taken by the LLM reasoning being crosschecked by the prediction MCP), (yellow: determined steps of the MCP server)
  • Figure 4: Representation of the Synthetic accessibility and synthetic complexity scores in OMG database.
  • Figure 5: Representation of Principal components of the latent space of the polymer SMILES space with colour indexing of the properties.
  • ...and 5 more figures