{
  "abstract": "Although three-dimensional flow-matching models have become powerful tools for efficient sampling of high-quality 3D molecular structures, reward-based reinforcement learning for flow-matching models is complicated by the absence of a tractable trajectory likelihood for hybrid molecular states comprising categorical features and continuous coordinates. Existing policy-optimization strategies address this difficulty by stochasticizing the generative process or constructing likelihood surrogates, introducing additional approximations. Here, we present ReCoFlow, a novel trajectorylikelihood-free reinforcement learning framework for reward-contrastive post-training of pretrained molecular flow-matching models. Within ReCoFlow, reward-favored molecules provide positive endpoint supervision for the flow-matching model, whereas reward-disfavored molecules are used to construct reference-guided negative targets. Additionally, an anchor loss to an exponential moving average (EMA) reference model constrains the drift of the fine-tuned model. In benchmarking studies including physicochemical property optimization, three-dimensional shape optimization, and pocket-conditioned 3D molecular generation, ReCoFlow consistently improved the performance of flow-matching models over reward-weighted and surrogate-ratio optimization schemes while maintaining high molecular validity. These results suggest that ReCoFlow could be an effective reinforcement-learning strategy for goal-directed molecular design.",
  "abstract_shown": true,
  "affiliations": [
    "School of Computer Science and Engineering",
    "Sun Yat-Sen University",
    "Guangzhou National Laboratory",
    "School of Pharmaceutical Sciences",
    "Guangzhou Medical University"
  ],
  "article_type": null,
  "arxiv": null,
  "author_total": 4,
  "authors": [
    {
      "affs": [
        0,
        1,
        2
      ],
      "name": "Bohao Li",
      "orcid": "0009-0004-0627-806X"
    },
    {
      "affs": [
        0,
        1
      ],
      "name": "Yuedong Yang"
    },
    {
      "affs": [
        2
      ],
      "name": "Mingyuan Xu"
    },
    {
      "affs": [
        2,
        3,
        4
      ],
      "name": "Hongming Chen"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/a4hx58pw898bb.bib",
  "cited_by": 0,
  "doi": "10.26434/chemrxiv.15008550/v1",
  "fulltext_url": null,
  "id": "a4hx58pw898bb",
  "issns": [],
  "landing_url": "https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008550/v1",
  "language": null,
  "licence": "CC-BY",
  "licence_label": "CC BY",
  "licence_source": "journal",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "openalex",
      "1703|Computational Drug Discovery Methods"
    ]
  ],
  "openalex": "W7212084319",
  "pdf_url": "https://chemrxiv.org/doi/pdf/10.26434/chemrxiv.15008550/v1",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-09",
  "publisher": "American Chemical Society (ACS)",
  "ris": "https://research.draconitic.com/paper/a4hx58pw898bb.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:17:19Z",
      "native_id": "10.26434/chemrxiv.15008550",
      "source": "chemrxiv",
      "url": "https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008550/v1"
    }
  ],
  "subjects": [
    {
      "field": "chemistry",
      "origin": "source:chemrxiv",
      "subfield": "general"
    }
  ],
  "title": "Contrastive Reward Based Reinforcement Learning for Goal Directed 3D Molecular Design",
  "title_text": "Contrastive Reward Based Reinforcement Learning for Goal Directed 3D Molecular Design",
  "type": "preprint",
  "updated_at": "2026-10-30T12:17:19Z",
  "url": "https://research.draconitic.com/paper/a4hx58pw898bb",
  "venue": "ChemRxiv",
  "versions": []
}