MMP2Mol: a matched molecular pairs-based framework for ligand-based de novo drug design.

MMP2Mol enhances ligand-based drug design by increasing molecular novelty to over 79%, outperforming baseline models across multiple therapeutic targets with limited data.

• Why it matters: Developing target-specific molecules is challenging when bioactivity data are scarce, limiting the ability to generate relevant compounds efficiently in early-stage drug discovery.
• What they did: The framework combines matched molecular pair analysis with a pretrained chemical language model, extracting target-specific transformations and fine-tuning the model to focus on relevant chemical space, evaluated on ten targets.
• The result: MMP2Mol achieves high validity and uniqueness, significantly boosting molecular novelty and maintaining diversity, thereby enabling more focused and reproducible candidate generation under data-limited conditions.