SmileyLlama: Modifying Large Language Models for Directed Chemical Space Exploration

TL;DR

SmileyLlama demonstrates how fine-tuned large language models can effectively explore chemical space for drug discovery, generating valid, novel molecules with desired properties.

Contribution

The paper introduces SmileyLlama, a fine-tuned LLM that integrates chemical language modeling with reinforcement learning for targeted molecule generation.

Findings

SmileyLlama outperforms pre-trained LLMs and chemical language models in molecule validity and novelty.

Using DPO improves adherence to prompts and property optimization.

Framework can be extended beyond drug discovery to other chemical and biological applications.

Abstract

We show that large language model (LLMs) can be transformed via supervised fine-tuning (SFT) of engineered prompts into SmileyLlama for exploring the chemical space of drug molecules. We benchmark SmileyLlama against pre-trained LLMs and chemical language models (CLM) trained from scratch for generating valid and novel drug-like molecules, and use direct preference optimization (DPO) to both improve SmileyLlama's adherence to a prompt and as part of the iMiner reinforcement learning framework to predict molecules with optimized 3D conformations and high binding affinity to drug targets. By training an LLM to speak directly as a CLM, while retaining most of its natural language capabilities, we show that we can reliably generate molecules with user-specified properties rather than acting only as a chatbot with knowledge of chemistry or as a virtual assistant. While SmileyLlama is geared toward drug discovery, the SFT/DPO/LLM framework can be extended to other chemical, biological, and materials applications.