{
  "abstract": "Structure-sensitive properties (SSPs), including activity cliffs and chirality-dependent properties, challenge molecular machine learning because small structural perturbations can cause abrupt property changes and invalidate smooth structure–property assumptions. Here, we present CAMF (Chirality- and Activity-cliff-aware Multimodal Framework), a task-adaptive framework that models SSPs through selective integration of complementary molecular evidence. To systematically evaluate this problem, we construct SSPBench, a benchmark spanning 77 conventional ADMET and physicochemical tasks together with activity-cliff and chirality-sensitive benchmarks. CAMF integrates molecular embeddings and expert-defined descriptors using random-forest-based feature selection and adaptive fusion, enabling property-specific prioritization of informative signals while reducing multimodal redundancy. Across ten baselines, CAMF achieves the best overall performance on SSP tasks, improving mean R2 by up to 29.5% on activity-cliff datasets and reducing MAE by up to 23.3% on 90 364 chiral molecules with TD-DFT-computed optical rotatory strengths. Ablation analyses show that these gains arise from task-adaptive multimodal integration rather than naive feature concatenation. More broadly, our results reveal that modality relevance is strongly task-dependent, with descriptors and 3D geometry becoming especially important in non-smooth property regimes. Case studies further support the interpretability and practical utility of CAMF in identifying activity-associated substructures and clinically relevant toxicity liabilities.",
  "abstract_shown": true,
  "affiliations": [
    "Faculty of Applied Sciences, Macao Polytechnic University Macao 999078 China",
    "State Key Laboratory of Public Big Data, College of Computer Science and Technology, Guizhou University Guizhou 550025 China",
    "Department of Computer Science and Engineering, Chinese University of Hong Kong 999077 Hong Kong"
  ],
  "article_type": "research article",
  "arxiv": null,
  "author_total": 14,
  "authors": [
    {
      "affs": [
        0
      ],
      "name": "Shaolong Lin"
    },
    {
      "affs": [
        0
      ],
      "name": "Silong Zhai"
    },
    {
      "affs": [
        0
      ],
      "name": "Shihang Wang",
      "orcid": "0000-0002-4714-6504"
    },
    {
      "affs": [
        0
      ],
      "name": "Xinke Zhan",
      "orcid": "0000-0002-1235-3220"
    },
    {
      "affs": [
        1
      ],
      "name": "Yuquan Li"
    },
    {
      "affs": [
        0
      ],
      "name": "Weihong Li"
    },
    {
      "affs": [
        0
      ],
      "name": "Li Qin"
    },
    {
      "affs": [
        0
      ],
      "name": "Lin Shi"
    },
    {
      "affs": [
        0
      ],
      "name": "Yanan Tian"
    },
    {
      "affs": [
        0
      ],
      "name": "Kai Xu"
    },
    {
      "affs": [
        0
      ],
      "name": "Kewei Zhou"
    },
    {
      "affs": [
        2
      ],
      "name": "Chunbin Gu"
    },
    {
      "affs": [
        0
      ],
      "name": "Huanxiang Liu",
      "orcid": "0000-0002-9284-3667"
    },
    {
      "affs": [
        0
      ],
      "name": "Xiaojun Yao",
      "orcid": "0000-0002-6972-2971"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/6nby0w23qeggy.bib",
  "cited_by": 0,
  "doi": "10.1039/d6sc02982e",
  "fulltext_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC13551429/",
  "id": "6nby0w23qeggy",
  "issns": [
    "2041-6520",
    "2041-6539"
  ],
  "landing_url": "https://doi.org/10.1039/d6sc02982e",
  "language": "en",
  "licence": "CC-BY-3.0",
  "licence_label": "CC BY 3.0",
  "licence_source": "article",
  "licence_url": "https://creativecommons.org/licenses/by/3.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "heading",
      "Chemistry"
    ],
    [
      "openalex",
      "1703|Computational Drug Discovery Methods"
    ]
  ],
  "openalex": "W7205015747",
  "pdf_url": "https://pubs.rsc.org/sc/article-pdf/doi/10.1039/D6SC02982E/13739386/d6sc02982e.pdf",
  "pmcid": "PMC13551429",
  "pmid": "42712854",
  "published": "2026-09-02",
  "publisher": "Royal Society of Chemistry",
  "ris": "https://research.draconitic.com/paper/6nby0w23qeggy.ris",
  "sources": [
    {
      "first_seen": "2026-10-31T12:23:45Z",
      "native_id": "PMC13551429",
      "source": "pmc",
      "url": "https://doi.org/10.1039/d6sc02982e"
    }
  ],
  "subjects": [
    {
      "field": "chemistry",
      "origin": "journal-openalex:1605|Supramolecular Chemistry and Complexes",
      "subfield": "organic"
    }
  ],
  "title": "Task-adaptive multimodal molecular representations for structure-sensitive property prediction",
  "title_text": "Task-adaptive multimodal molecular representations for structure-sensitive property prediction",
  "type": "peer-reviewed",
  "updated_at": "2026-11-01T12:20:53Z",
  "url": "https://research.draconitic.com/paper/6nby0w23qeggy",
  "venue": "Chemical Science",
  "versions": []
}