MolX: Enhancing Large Language Models for Molecular Understanding With A Multi-Modal Extension
This work addresses the limitation of LLMs in chemistry by providing a more effective way to handle molecular data, which is incremental as it builds on existing LLM capabilities with a specialized module.
The paper tackles the problem of enhancing large language models (LLMs) for molecular understanding by introducing MolX, a multi-modal extension that uses encoders for SMILES strings and 2D graphs, along with molecular fingerprints, to improve performance on tasks like molecule-to-text translation and property prediction, achieving better results than baselines with only 0.53% to 0.82% additional trainable parameters.
Large Language Models (LLMs) with their strong task-handling capabilities have shown remarkable advancements across a spectrum of fields, moving beyond natural language understanding. However, their proficiency within the chemistry domain remains restricted, especially in solving molecule-related tasks. This challenge is attributed to their inherent limitations in comprehending molecules using only common textual representations, i.e. SMILES strings. In this study, we seek to enhance the ability of LLMs to comprehend molecules by equipping them with a multi-modal external module, termed MolX. Instead of directly using SMILES strings to represent a molecule, we utilize specific encoders to extract fine-grained features from both SMILES string and 2D molecular graph representations for feeding into an LLM. A hand-crafted molecular fingerprint is incorporated to leverage its embedded domain knowledge. To establish an alignment between MolX and the LLM's textual input space, the model in which the LLM is frozen, is pre-trained with a strategy including a diverse set of tasks. Experimental evaluations show that our proposed method outperforms baselines across downstream molecule-related tasks ranging from molecule-to-text translation to molecular property prediction, with and without fine-tuning the LLM, while only introducing a small number of trainable parameters-0.53% and 0.82%, respectively.