SupraBench: New Benchmark Evaluates AI Language Models on Supramolecular Chemistry Tasks
Researchers have released SupraBench, the first systematic benchmark to evaluate large language models (LLMs) on supramolecular chemistry tasks such as binding affinity prediction and host-guest system design. The benchmark includes four fundamental chemistry tasks plus a vision-based molecular identification task, along with a 16-million-token corpus of supramolecular chemistry literature. The work reveals that current LLMs have significant room for improvement in chemistry reasoning, with distinct failure modes across different task types.
SupraBench is a new benchmark designed to systematically evaluate how well large language models perform on supramolecular chemistry tasks—a field focused on non-covalent host-guest assemblies with applications across multiple domains. The benchmark, developed in collaboration with domain experts, includes four core tasks: binding affinity prediction, top-binder selection, solvent identification, and host-guest description, plus an auxiliary vision-based task for molecular identification. Researchers also released SupraPMC, a curated 16-million-token corpus of supramolecular chemistry articles extracted from Europe PMC to support domain-specific model adaptation. Testing across a broad range of open and proprietary LLMs revealed substantial performance gaps across all tasks. The study found that while domain adaptation pretraining using SupraPMC improved in-distribution regression performance, it created trade-offs with strict output formatting requirements, and different task families showed distinct difficulty profiles and failure modes.
What's missing
The paper does not specify which proprietary and open-source LLMs were benchmarked, making it difficult to assess relative performance across specific models. Additionally, the practical implications for accelerating supramolecular chemistry research timelines and cost savings compared to traditional dry-lab verification are not quantified.
What different sources said
- arXiv cs.AICenter
SupraBench: A Benchmark for Supramolecular Chemistry
Related
Topology-Aware Thermodynamics Improves DNA Probe Specificity Design
Researchers developed a new framework for designing DNA probes that accounts for the spatial organization of matched sequences, not just overall thermodynamic stability. Traditional methods rely on scalar measures like melting temperature and free energy, which miss how mismatches are distributed along the probe. The approach could improve diagnostic accuracy in applications like HPV detection and gene expression profiling.
Study Identifies Optimal Thermal Dose for Combining Focused Ultrasound with Immunotherapy in Tumors
Researchers used multimodal PET imaging to identify an optimal thermal dose range for focused ultrasound ablation that destroys tumor tissue while preserving conditions for immunotherapy delivery. The study found that excessive heating collapses blood vessels needed for antibody access, while insufficient heating fails to adequately reduce tumor burden. The findings could guide clinical design of combination treatments pairing thermal ablation with immunotherapies.
Plant MSH1 Protein Functions as Mismatch-Directed Nuclease for Organelle Genome Maintenance
Researchers have identified the precise mechanism by which the AtMSH1 protein in Arabidopsis plants recognizes and cleaves DNA mismatches and lesions, preventing mutations in organellar genomes. The protein combines a DNA mismatch recognition module with a nuclease domain that makes staggered cuts at specific positions relative to DNA damage. This discovery explains how plants maintain unusually low mutation rates in their mitochondrial and chloroplast DNA compared to other eukaryotes.