Do LLMs Truly Generalize in the Molecular Domain? A Perturbation-Based Analysis

• URL: http://arxiv.org/abs/2607.01800v1
• Date: Thu, 02 Jul 2026 07:16:28 GMT
• Title: Do LLMs Truly Generalize in the Molecular Domain? A Perturbation-Based Analysis
• Abstract summary: We introduce a Molecular Perturbation framework that generates structural variants of training molecules under controlled Graph Edit Distance (GED)<n>Our analysis shows that even a single edit can cause substantial performance drops on common molecular tasks, revealing a narrow local trust region and fragile sensitivity to structural changes.<n>Our experiments also examine whether ICT confers robustness under controlled structural perturbations.
• Score: 35.63799709687231
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Large Language Models (LLMs) have recently shown promise in molecular discovery, yet a gap remains between their probabilistic nature over discrete sequential tokens and the rigid topological constraints of chemical space. This raises the question of whether molecular LLMs can generalize beyond the local neighborhoods induced by their sequence-based representations. To systematically investigate this question, we introduce a Molecular Perturbation framework that generates syntax-valid structural variants of training molecules under controlled Graph Edit Distance (GED) to probe the manifold regularity of molecular LLMs. Our analysis shows that even a single edit can cause substantial performance drops on common molecular tasks, revealing a narrow local trust region and fragile sensitivity to structural changes. Since similar molecules tend to exhibit similar properties, In-Context Tuning (ICT), which anchors predictions on structurally similar molecules, offers a natural way to mitigate such fragility. Our experiments also examine whether ICT confers robustness under controlled structural perturbations, and the results suggest that it can partially expand the local trust region and offer a promising direction for stabilizing molecular LLMs against structural variation.

Related papers

• Structural Hierarchy and Geometry in Molecular Representation Learning [0.0]
  We study whether explicitly encoding a molecule's Bemis-Murcko scaffold and using it to supervise the molecular embedding changes what the model learns.<n>Across two augmentation strengths, scaffold-supervised models consistently organize molecules according to both identical and structurally related scaffolds.
  arXiv  Detail & Related papers  (2026-08-30T16:34:17Z)
• MolLingo: Molecule-Native Representations for LLM-Powered Scientific Agents [57.28652712579434]
  We present MolLingo, a multi-agent system that emulates the reasoning process of a chemist to automate molecular design.<n>MolLingo addresses this by coordinating a Literature Agent, a Chemist Agent, and an Orchestrator through a shared memory module, with each agent equipped with domain-specific tools.<n>As a case study in early-stage therapeutic design, MolLingo grounds the Chemist Agent's reasoning in binding site geometry and residue-level protein context.
  arXiv  Detail & Related papers  (2026-05-27T02:11:23Z)
• Molecular Representations in Implicit Functional Space via Hyper-Networks [53.70982267248536]
  We argue that molecular learning can instead be formulated as learning in function space.<n>We instantiate this formulation with MolField, a hyper-network-based framework that learns distributions over molecular fields.<n>Our results show that treating molecules as continuous functions fundamentally changes how molecular representations generalize across tasks.
  arXiv  Detail & Related papers  (2026-01-29T21:13:37Z)
• How well can off-the-shelf LLMs elucidate molecular structures from mass spectra using chain-of-thought reasoning? [51.286853421822705]
  Large language models (LLMs) have shown promise for reasoning-intensive scientific tasks, but their capability for chemical interpretation is still unclear.<n>We introduce a Chain-of-Thought (CoT) prompting framework and benchmark that evaluate how LLMs reason about mass spectral data to predict molecular structures.<n>Our evaluation across metrics of SMILES validity, formula consistency, and structural similarity reveals that while LLMs can produce syntactically valid and partially plausible structures, they fail to achieve chemical accuracy or link reasoning to correct molecular predictions.
  arXiv  Detail & Related papers  (2026-01-09T20:08:42Z)
• $\text{M}^{2}$LLM: Multi-view Molecular Representation Learning with Large Language Models [59.125833618091846]
  We propose a multi-view framework that integrates three perspectives: the molecular structure view, the molecular task view, and the molecular rules view.<n>Experiments demonstrate that $textM2$LLM achieves state-of-the-art performance on multiple benchmarks across classification and regression tasks.
  arXiv  Detail & Related papers  (2025-08-12T05:46:47Z)
• Structural Reasoning Improves Molecular Understanding of LLM [18.532188836688928]
  We show that large language models (LLMs) still struggle to reason using molecular structural information.<n>We propose an approach that sketches molecular structures for reasoning.<n>We present two frameworks for scenarios where the target molecule is known or unknown.
  arXiv  Detail & Related papers  (2024-10-08T01:49:48Z)
• LDMol: A Text-to-Molecule Diffusion Model with Structurally Informative Latent Space Surpasses AR Models [55.5427001668863]
  We present a novel latent diffusion model dubbed LDMol for text-conditioned molecule generation.<n> Experiments show that LDMol outperforms the existing autoregressive baselines on the text-to-molecule generation benchmark.<n>We show that LDMol can be applied to downstream tasks such as molecule-to-text retrieval and text-guided molecule editing.
  arXiv  Detail & Related papers  (2024-05-28T04:59:13Z)
• Multi-channel learning for integrating structural hierarchies into context-dependent molecular representation [10.025809630976065]
  This paper introduces a novel pre-training framework that learns robust and generalizable chemical knowledge.<n>Our approach demonstrates competitive performance across various molecular property benchmarks.
  arXiv  Detail & Related papers  (2023-11-05T23:47:52Z)
• Towards Predicting Equilibrium Distributions for Molecular Systems with Deep Learning [60.02391969049972]
  We introduce a novel deep learning framework, called Distributional Graphormer (DiG), in an attempt to predict the equilibrium distribution of molecular systems. DiG employs deep neural networks to transform a simple distribution towards the equilibrium distribution, conditioned on a descriptor of a molecular system.
  arXiv  Detail & Related papers  (2023-06-08T17:12:08Z)
• Domain-Agnostic Molecular Generation with Chemical Feedback [44.063584808910896]
  MolGen is a pre-trained molecular language model tailored specifically for molecule generation. It internalizes structural and grammatical insights through the reconstruction of over 100 million molecular SELFIES. Our chemical feedback paradigm steers the model away from molecular hallucinations, ensuring alignment between the model's estimated probabilities and real-world chemical preferences.
  arXiv  Detail & Related papers  (2023-01-26T17:52:56Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.