pharmacology

Introduces MARD, a 7B-parameter model for mechanism-level drug-drug interaction prediction using mirror-augmented reasoning distillation, achieving state-of-the-art accuracy at ~1% of frontier API cost and demonstrating genuine pharmacological reasoning over memorization.

This paper investigates how LLMs rely on morphological cues (affixes) to make pharmacological inferences, demonstrating that models can confidently generate plausible content for fictitious drug names based solely on affix heuristics, which poses a subtle safety risk.

This paper formalizes transcriptome-based drug design (TBDD) as a generative inverse problem and proposes CURE, a multi-resolution transcriptome-guided diffusion framework that generates drug molecules conditioned on desired transcriptomic state transitions.